This page is where I will share some of my thoughts, feelings, and experiences dealing with computers. I always try to do a good job of documenting what I've done to fix my computer woes. I've done this online to hopefully help others with computer problems similar to mine. I don't plan to translate this page into Esperanto.
Ĉi tiu paĝo estas kie mi distribuos iun el miaj pensoj, sentoj, kaj spertoj pri komputiloj. Mi ĉiam penas dokumenti bone kion mi faris ripari miajn komputilajn problemojn. Mi faris ĉi tio sur la reto por helpi aliajn kiuj havis similajn komputilajn problemojn. Mi ne planas traduki ĉi tiun paĝojn al Esperanto.
machine | type | processor | OS | disk | mem | location |
---|---|---|---|---|---|---|
pulsar | shuttle XPC | AMD XXXXXX | ubuntu 11.04 | 250 G | 2.0 G | home |
brebis | MacBook Pro 5.3 | Intel Core 2 Duo T9550 2.8 GHz | Mac OS X 10.6.6 | 480 G SSD, 500 G | 4.0 G | home |
emmental | VM in brebis | 1 virtual x86 | ubuntu 10.04 | 90 G | 1.0 G | home |
camembert | VM in brebis | 1 virtual x86 | Windows XP SP3 | 30 G | 1.0 G | home |
maredsous | Dell Precision T3500 | Intel Xeon W3503 2.4 GHz | Ubuntu 9.10 | 2 T | 4.0 G | numediart |
mozzarella | HP blahhlah | 2003 technology | Ubuntu 9.10 | 120 G | 2.0 G | home - Belgium |
parmesan | Dell Inspiron 8200 | Intel Pentium 4 Mobile 1.6 GHz | ubuntu 9.10 | 80 G | 1.0 G | home - Belgium |
chèvre | Dell Latitude D830 (laptop) | Intel Core 2 Duo T9300 2.5 GHz | ubuntu 9.04 | 500 G | 2.0 G | STOLEN |
reblochon | MacBook Pro 5.1 | Intel Core 2 Duo T9550 2.67 GHz | Mac OS X 10.5.6 | 300 G | 4.0 G | STOLEN |
maroilles | VM in reblochon | Intel Core 2 Duo T9550 2.67 GHz | ubuntu 9.04 | 40 G | 1.92 G | STOLEN |
sedusa | frankenputer desktop | AMD Athlon 2000+ 1.67GHz | ubuntu 7.10 | 80G 250G | 1.25G | Pink Palace (communal) |
pinkmame | frankenputer desktop | AMD Athlon XP 2600+ | debian 4.0 etch | 500G | 512M | Pink Palace (MAME) |
ricotta | Apple iMac G3 | PowerPC G3 | ubuntu 7.04 | 80G | 512M | home (stereo) |
bleu | IBM Thinkpad R50 (laptop) | Intel Pentium M 1.6GHz | ubuntu 7.10 | 40G | 1.25G | home (broken screen) |
lighthill | Dell Precision 530 | Dual Intel Xeon 1.80GHz | ubuntu 7.10 | 80G 250G | 1.00G | research lab |
fenestro (etc4-156-3) |
Dell Precision 530 | Dual Intel Xeon 1.80GHz | ubuntu 7.10 | 80G 80G | 1.00G | research lab (window) |
gouda | Dell Inspiron 7000 laptop | Intel PII 350MHz | ubuntu 6.06 | 6G ?? | 384M ?? | home (stereo) |
pimento | frankenputer desktop | AMD Athlon XP 2000+ 1.67GHz | debian sarge Windows ME |
60G 12G | 256M | home |
I ran into the situation where key repeat wouldn't work in TextEdit (mac), because a "helpful" international keys typer kept popping up instead of keys repeating. I remembered having run into this situation before. After doing some digging, I refound this helpful page about the problem.
TL;DR -- The following command in a terminal fixes it.
defaults write -g ApplePressAndHoldEnabled -bool false
Lollipop installed itself to my Nexus 4, and generally it's been a good experience. My WiFi started working better, and the battery seems to last longer. However, most of my custom-made notification sounds wouldn't play any sound, which was really annoying. I think I finally stumbled upon the cure. Basically, many of the sounds were under 0.1 seconds. I just added 0.1 sec of silence to both ends of the sounds, and those all worked. I both originally created and recently edited the sounds with Audacity.
After getting a few "Permission denied" errors when trying to remove files and directories which I apparently had the correct permission to remove from the command line with
rm -rf blah
I figured out, after reading a few web pages that there were some ACLs (access control lists) associated with files and directories in os x. This meant that I had to remove some of those before I'd be able to remove the files. I can see these ACLs with
/bin/ls -le
And then I could remove all of the ACLs with
sudo /bin/chmod -N *
Yay! Now
rm -rf blah
works!
grep -rl string1 somedir | xargs sed -i 's/string1/string2/g'
Note that this presumes non-os-x sed. On OS X, either use gsed (gnu sed from either brew or macports), or add a couple of quotes in the sed line like this.
grep -rl string1 somedir | xargs sed -i "" 's/string1/string2/g'
Deleting files from a mounted sparsebundle will not reduce the size of the sparsebundle. In order to do that, you must compact the sparsebundle itself.
hdiutil compact blah.sparsebundle
Apparently OS X/HFS+ has extra flags that can prevent files from being changed. To see these extra flags, you have to use the OS X version of ls and do the following.
/bin/ls -lOe /path/to/item
If the uchg flag is set, then you have to do the following.
chflags nouchg /path/to/item
I found the info at this forum page.
Of course, there's an Apple Support Page about how to make it work.
Wow, I was totally surprised to discover that OS X is running CUPS as the printer server, and setting up a RAW printer queue is almost the same as it is on linux. There is an extra "Add Class" step that has to happen, but it's all detailed on jzebra's site.
I followed another web page to do the upgrade, but basically, it was
cd /Library/Java/Extensions; rm
libsvnjavahl-1.jnilib && ln -s libsvnjavah1.jnilib libsvnjavahl-1.jnilib
)svn upgrade
and it all worked. Yay! Supposedly this is *not* going to mess with the repository at all. We shall see.
All of this was following John Hawksley's "Note To Self" blog entry called Eclipse Juno - Update Hurdles.
My backup scheme is pretty simple and complex at the same time. I'm sure that there are better/easier schemes, but most of them seem to involve someone else having access to my data. i'm fine with keeping track of this stuff myself for now, though this doesn't go along with the "cloud" mentality.
Basically, I wrote a script which uses rsync to put a copy of my data into a directory mounted on my home directory. This directory can be an sshfs mount of a directory on another machine, or a mounted directory on some USB device. At this point, the USB devices all have encrypted sparsebundles on them, so that there's no way to get to the data on the USB device. Yay! This does mean that if I get a new USB device that I'd like to store data on like I did today, I'll need to create a new encrypted sparsebundle.
And that should be it. Easy Peasy. For me, I then have to add an entry for the encrypted bundle on the new USB device into my directory mounting program, mySAUVEGARDE, and then use that to mount to the backup directory. Yay!
Oh, my dog, I love synergy! It's awesome. But, unfortunately, there are some problems with the OS X app interface that keep synergy from working well with double clicks with OS X as a client. This ends up meaning that I needed to use a backported version of synergy to get double clicks. It's available at http://synergy-foss.org/pm/attachments/2868/synergy-plus-1.3.5rc_-Darwin-Universal.dmg.
It turns out that you can see the contents of the .pkg files that some programs come in for OS X using the following commands.
First, determine the name of the package according to the pkgutil program.
pkgutil --pkgs
You might want to add a piped grep to that to find the pkg you're interested in. Then
pkgutil --files packagename
Now you have a list of the files created/used by the pkg.
I wanted to turn it off. Here's the command line.
defaults write com.apple.dock workspaces-swoosh-animation-off -bool YES && killall Dock
To turn it back on.
defaults delete com.apple.dock workspaces-swoosh-animation-off && killall Dock
I recently bought an HP 2310m monitor. It's 16:9, and though it has an picture, it wobbles on it's stand quite a bit. I think it's basically due to it's ridiculously wide aspect ratio, and the torque the monitor generates against its stand. I've been trying to figure out how to fix this without spending any money, when today, I realized that I could put a shelf next to it and have it bump up against the shelf. Yay! This is a lotech solution to an annoying problem.
bash name completion (tab after typing some of the characters on the command line) wasn't working for me on OS X. Basically, when I was hitting tab to complete the name of a symlinked directory, the slash afterwards was being left out. This is actually reasonable behavior, but it turns out it's controllable, too. I found a page which suggested putting
set mark-symlinked-directories on
into ~/.inputrc. I did that, logged out and back in, and voila, symlinks tab complete with slashes at the end.
Okay, so I switched to using gnu's ls instead of the BSD version of ls, because I like my files colored by extension as well as type (dir, link, etc). Basically it just takes a
sudo port install coreutils +with_default_namesand a subsequent change to your $PATH which is described upon running that command.
The colors aren't necessarily beautiful, especially if you've installed some custom colors for whatever terminal program you're using like I have. I'd figured out how to set some of the colors, but it wasn't until I ran
dircolors --print-database
that I really figured out what could be set.
Okay, I've made a huge decision. I'm switching over to OS X as my main computer these days. There are lots of reasons for the switch, but the main one was that now that my email is not on my local machine, running a virtual Linux box just doesn't make sense anymore. I'm ready to catch up with the rest of the modern world with respect to computing. I actually made the switch on Monday the 14th, and have been catching up with only minor problems here and there ever since.
The users' home directories on OS X are in /Users, but there actually is a /home directory. This is used by automountd to automatically mount some NFS partition if you want to use it. That's probably not going to happen on my home machine, so I've turned that off by commenting out the /home line in /etc/auto_master and rebooting.
The reason I did this was to link /Users to /home so that my duplicity script that backs up to konfuzo.net will work without any hitches. I made the link and the duplicity script seems to work great!
There isn't a good way to set the default gnome-terminal size in Lucid Lynx, so I ended up changing /usr/share/vte/termcap/xterm and set the "li" value from 24 to 39. That works, and is okay since there are no other users on my system.
I found a trick to REALLY increase my Vibrant's battery life. Set the background to black. The Samsung Vibrant uses AMOLED (active matrix organic LED) technology, so the screen actually turns on when the screen is lit. There is no common backlight. This means that not turning on LEDs (setting them to black) actually uses less power, and therefore less battery. Calculations indicate a 300% battery life increase. YAY!
By default, the cursor in gnome-terminal blinks. Fortunately, Mr. Google knew the solution.
gconftool-2 -s /apps/gnome-terminal/profiles/Default/cursor_blink_mode -t string off
I got a smartphone. I got a Samsung Vibrant. It's fun!
I wanted a custom default ringtone. I found a page that talks about how to make that happen.
http://forum.xda-developers.com/showthread.php?t=736109
Basically, one puts mp3s in the following directories, creating the directories as necessary.
/sd/Media/Audio/Ringtones /sd/Media/Audio/Notifications /sd/Media/Audio/Alarms
Okay, there are some serious problems with just using ntp, unfortunately. Basically the virtual machine gets "paused" when the laptop is closed, so it doesn't know that it needs to look up a new time when it's reopened. On reopening the laptop, I have to use ntpdate every time to reset the time correctly. This seems like a no-win situation. Feh.
VBoxManage setextradata "guestname" "VBoxInternal/Devices/VMMDev/0/Config/GetHostTimeDisabled" "1"
At this point, I've decided to turn of time synchronization between the guest and host. This is achieved on an OS X host with
VBoxManage setextradata "guestname" "VBoxInternal/Devices/VMMDev/0/Config/GetHostTimeDisabled" "1"
replacing "guestname" as appropriate. I then went and reinstalled ntp on the guest. Hopefully this will keep things synched. We'll see.
Okay, so my ubuntu guest on virtual box running on an OS X host has consistently lost around 20 minutes and only catches up after a period of major downtime with the screen closed. Annoying!
I found a page, http://wiki.category5.tv/Correct_VirtualBox_Time_Sync, that describes techniques used to fix the problem, and I'm going through them in the order they suggest, but I'm guessing in the end that I'll need to disable the VirtualBox time syncing and just use ntp. Here's what I've tried so far.
EDIT 2010-09-13 I just saw that these are things to do for linux hosts, not linux guests. I think the answer is going to end up being turn off timesyncing for the linux guest and use ntp. Feh.
I don't know why, but on one of the updates to VirtualBox, sound stopped working on my linux host (ubuntu 10.04 Lucid Lynx). The fix was to stop the virtual host, rechoose CoreAudio in the Sound options for the guest and restart. Simple, but why?
I love using Terminal!
drutil status diskutil unmount /dev/disk2 dd if=/dev/disk2 of=whatever.iso bs=2048
openssh-server keychain fvwm m4 mutt fetchmail procmail vim gcal apache2 php5 libapache2-mod-php5 imagemagick mplayer ntp mkisofs aumix wodim cdrecord synergy apt-file sun-java6-jre sun-java6-plugin qiv unsort randomize-lines bittornado flac dbus-x11 subversion mpg321 wmbattery ubuntu-restricted-extras gnucash envyng-core figlet build-essential spamassassin audacity csound nfs-common enscript paps tree texlive texlive-latex-extra texlive-humanities ispell lame console-data grip openoffice.org gnugo qgo sanduhr manpages-dev abook remind eclipse git-core duplicity sshfs encfs exim4 xscreensaver xscreensaver-data-extra xscreensaver-gl-extra bsdgames nethack-console fortune-mod fortunes fortunes-fr fortunes-eo
sudo dpkg --force-architecture -i AdobeReader_enu-8.1.3-1.i386.deb
options snd-intel8x0 ac97_clock=48000 options snd slots=snd-intel8x0 alias snd-card-0 snd-intel8x0
Just putting this here for reference: irc.epiknet.org.
I needed to add this to .mailcap.
text/html; firefox -remote 'openURL(%s)'
I installed a new kernel in the ubuntu updates today, and the screen and mouse started behaving weirdly. After a short time, I realized that I needed to install the Virtual Box Guest Additions.
To install those, I chose install Guess Additions from the Virtual Box Devices menu while running the guest. It took me a minute to realize that I also needed to
mount /dev/cdrom /media/cdrom
but once I did that, everything else went smoothly.
Okay, so I'm now silly paranoid about data sitting on my machines, so I went and got rid of the virtual machine which had an unencrypted version of my home directory sitting in it. The disk image was sitting in /Users/nodog/Library/VirtualBox/HardDisks on my mac, so now even someone stealing this computer couldn't create a new virtual machine and attache that disk image to it.
As far as I can remember, there are no unencrypted versions of my home directory anywhere now. There are, however, a bunch of encrypted copies sitting all over the place. I'm pretty happy about that.
Speaking of encrypted home directories, I did think a bit about where the encryption should actually sit, especially because I have several options with the virtual machines. I could either encrypt the virtual files in the virtual machine, or encrypt the file on the host machine which contains the virtual hard disk image. Either works, although I feel like the latter gives a bit more room for a would-be hacker to work with. Also, if someone steals my computer when it's on, then it feels like even more room, because they can crack either screensaver and they've got the data.
Basically, I decided that the encryption should be at the level immediately next to the data, so I'm encrypting the home directory inside the virtual machine. I'm leaving the host system unencrypted for now and making sure that I do not store anything important there. This also means that I don't log into any important websites and store passwords. Can do.
If something were to go terribly wrong, I do want to have access to all of my copies of my encrypted home directory. I want to write down here how to get at it.
To access the stuff in encfs on the external drives, just do the following
mkdir myTempEnc mkdir myTemp sudo mount /dev/sdb1 myTempEnc encfs `pwd`/myTempEnc `pwd`/myTemp
To access the stuff in encfs on other computers, just do the following
mkdir myTempEnc mkdir myTemp sshfs -o uid=1000,gid=1000,idmap=user otherhost:remoteEncFSDir myTempEnc encfs `pwd`/myTempEnc `pwd`/myTemp
For the Duplicity backups on other computers, to restore my entire home directory to some directory
mkdir myTemp duplicity scp://me@otherHost/duplicityDir myTemp
To restore just one or a few files
mkdir myTemp duplicity --file-to-restore some/file scp://me@otherHost/duplicityDir myTemp
The duplicity stuff may take some fiddling, but I remember testing it, and remember not fiddling much if any.
I don't want iTunes on my machine if I can help it. I found a site to help me remove it, but then Software Update wants me to install the upgrade all the time. I finally found the Update > Ignore Update menu item today. Yay! Hopefully Software update won't bother me again.
I got a hint on how to do this task from this page. They also suggest that it might be good to use a USB stick to store the my_packages.txt file and /etc/apt/sources.list as well.
sudo dpkg --get-selections "*" > /media/disk/my_packages.txt
sudo dpkg --set-selections < /media/disk/my_packages.txt sudo apt-get -u dselect-upgrade
I installed skype from the medibuntu repositories, and to my surprise, it failed on initial start up. After trying in a terminal and getting the very useful message of "Aborted". I removed ~/.Skype/shared.* as described on an ubuntu forum page and skype started working. Feh.
I haven't been able to come up with a viable solution to the problems with utf8 filenames being stored differently on Mac OS X (really HFS+) and Linux (really ext2/ext3/ext4) on a shared directory in VirtualBox. The basic problem is that OS X stores the utf8 characters in a format known as MAC-UTF8, and when one rsyncs or copies these files with a Linux system, there is all sorts of not understanding the differently encoded names as being the same file.
The only solution that I could come up with that doesn't involve network traffic is to create a disk image file, format it with the ext2/ext3 filesystem, and then mount in read-write on the Linux system and read-only on the OS X system. This is a bit non-optimal, in that the disk image can't be one of those cool grow-as-you-go disk images since those depend on the local OS to make those work. I'll be wasting some space, that's for sure, but the compatibility issues are worth it.
The steps are:
hdiutil create -size 300g -type UDIF /OSXpathname/to/diskimagefile
mke2fs -j /linuxpathname/to/diskimagefile.dmg
sudo mount -o loop /linuxpathname/to/diskimagefile.dmg /my/mount/point
hdiutil attach /OSXpathname/to/diskimagefile
Just want to keep track of how I generated the French locale.
locale-gen fr_FR.UTF-8 locale-gen fr_BE.UTF-8
Well, that's funny. It makes sense that if I have an encrypted home directory on my machine, I would want to mount it with 700 permissions. That prevents me from accessing my public_html folders in my home directory, unfortunately. For the moment, I've just made my home directory 755, but this may change on the next mount of the home directory. I'll have to wait and see. (or test!)
In order to download all of my LJ entries to my local computer, I insalled charm, a commandline blogging program. It works well, but required one very strange thing in order to install it. I had to download the source and run a setup.py script as root. It's because it has a specific python module to load, but that was weird to see in a ubuntu package.
Since I run fvwm instead of gnome, I've often wondered if a reboot was required after an update/upgrade. Well, it turns out that the file /var/run/reboot-required is created during an upgrade if, in fact, a reboot is required. I can add a little call to this in my ubuntuupgrade alias.
I've bemoaned the lack of command completion after sudo for many years, but just lived with it. How foolish of me! It turns out that bash has much better command completion available. To enable it systemwide, just uncomment the if/fi lines in /etc/bash.bashrc dealing with "bash_completion". Cool!
Somehow, I managed to munge a file, and I was able to restore it using duplicity. I was even able to check when it got munged by using the -t time specification on the restore. Duplicity rocks.
I did find out through this test, however, that duplicity will not overwrite existing files, so this means that I cannot use it as a synchroniser between systems. Ah well. At the moment, I only have one system, so that's not a big deal, but someday, I will have more than one computer again, and at that point, I will want to be able to synch between computers. myRsync on to a encfs will stay around.
Okay, so the Terminal/bash in OS X tricked me with strange filename completion around UTF8 characters. Basically, I have a huge amount of files in a directory structure archive and some of them include UTF8 characters in the filenames, which is completely appropriate for those files. It seems that OS X Snow Leopard seems to have done a pretty good job of handling the weirdness that is HPFS+ UTF8 encoding, but when I would try to do filename completion on those files both through Terminal and through a shared folder in Virtualbox, the filenames would not complete. Well it turns out that if you have a file named "Señor" if you try to do filename completion on "Señ" it won't complete. Because of the weird way that HPFS+ stores the UTF8 characters as separate letter + diacritical, one needs to try to do autocompletion on "Sen". What a pain!
I do remember that I was still having troubles with the linux-OS_X UTF8 encoding difference on one of the transfers that I was doing with this archive. I'll have to keep an eye out for that still.
I used the OS X program Ukelele to create some custom keyboard layouts so I can type in Dvorak on the Belgian keyboard. It works pretty well. Tonight, I added dead keys so that I could type in OS X the same way that I type in Linux. It's not perfect, but it works. I still need to go back and fix the dead key for the double dots and the ŭ.
Ubuntu decided to make a "throbber" that goes back and forth across a small part of the screen. I do not want! It turns out that it's relatively easy to comment out the following lines from both /etc/gdm/Init/Default and /etc/gdm/PreSession/Default to remove it.
if [ -x '/usr/bin/xsplash' ]; then /usr/bin/xsplash --daemon fi
In looking around for encrypted file systems, I found the loopback filesystem, but it kind of sucks because one has to specify the size of the final file container ahead of time. One can cause them to grow, but there just seems to be a lot of wasted space in this solution.
The fix for this seems to be the EncFS filesystem which allows the creation of an encrypted filesystem inside a directory on the local filesystem. With sshfs in place, this could also be a directory on a remote filesystem. Well, it seems thot encfs over sshfs is just too slow over a slowish network to really be usable for me in the way that I want. (I expect this would also be true for encrypted loopback filesystems as well, but I don't know.) There is most certainly a bottleneck somewhere, and I think it's the network, unfortunately. I expect that I will be using encfs for local disks, but not for remote systems unfortunately.
Speaking of my needs, I have two basic mechanisms that I want. One is an encrypted backup of large directories (one ~30G and the other ~300G) on the network. The other is syncronization of those same directories locally. I'm still trying to decide if I want the local disks to be encrypted or not.
I've now tried a few different mechanism for the encrypted backups of large directories with all of this, and the one that I like the best is duplicity for remote encrypted incremental backups. This isn't a total solution to my backup desires, but it is a solution that I'm liking more and more. (Hunh. It just finished a 3.3G backup to another machine on the LAN in 26 min.)
I have a home-rolled solution for the synchronization using rsync, but my process here.
to add a user to a group
dscl . -append /Groups/operator GroupMembership username
Before my computers were stolen, I was enjoying learning French by playing Neverwinter Nights in French on my Linux box. Well, now, I have my work Mac and decided to continue the process, but getting everything installed to play in French was tricky. First, I installed the Diamond Edition of NWN (patch 1.69) that I had in English. This was followed by extracting the French Language Files v1.29 from the Bioware Linux client page. I followed this by installing the French language patch 1.68 from the Bioware Patches and Updates page. Everything seems to be working well so far, but I'll probably need to repeat the process to install the French language packs for the expansion packs, too.
UPDATE 2009-12-25: I had to copy the Data_Shared.zip, Language_data.zip, and Language_update.zip files from the SoU disk before I could get the French spoken (actually, I only think that Data_Shared.zip mattered, but good to be complete). I then installed the latest mac French patch afterwards.
My new MacBook Pro 5.3, brebis, which was graciously bought for my by my lab, and my old MacBook Pro 5.1, reblochon, both have AZERTY (Belgian/French) keyboards. This is mostly fine, since I overlay the Dvorak keyboard on it anyway, but the key which normally types "`" and "~" now types "§" and "±". "`" and "~" are now typed by an extra key (labelled "<" and ">") to the right of the left shift key. This was just too hard to get used to, so I've used the software Ukelele to create a custom keylayout to turn off the key next to the left shift key and move the typing of "`" and "~" to where I want it. Just put nodog-Belgian-Dvorak.keylayout into /Library/Keyboard Layouts, log out, and log back in. It should be available in System Preferences > Language and Text > Input Sources in alphabetical order.
I forgot that I also had to do that key switching in Ubuntu as well. I wrote a short script, asdf, which sets up the good keyboard and then fixes the two problem keys.
#!/bin/bash setxkbmap us dvorak -option compose:ralt if [ emmental = $HOSTNAME ]; then xmodmap -e "keycode 94 = grave asciitilde" -e "keycode 49 =" fi
Okay, so now that I understand what can happen when one loses all files and disks, I want a remote backup again. The problem with remote backups before was that I didn't want my data sitting unencrypted on a remote machine. I think I have found the solution, and it will work well with my already created backup system.
It involves two mechanisms working in parallel. First is sshfs and the other is an encrypted loopback filesystem. sshfs allows me to mount a remote directory on my local machine very simply. The fact that it does this encrypted is great but not actually necessary in this case. The encrypted loopback filesystem allows me to mount a single file as if it were a filesystem with the extra feature that the file is encrypted. This is more important to me. According to what I've read, this file can also be grown after the fact, which is pretty cool.
Here are two pages I've found about it, and I'll be going through these steps soon.
Well, basically, all USB drives don't work well with the Snow Leopard host for VirtualBox guests. I've got a temporary working solution for one drive (a Western Digital "My Passport Elite"), but I can't get a different drive (a Western Digital "My Passport Essential") working at all. Bummer. It's in the forums, but no one's found a solution yet. I beat my head against this all day, basically.
openssh-server keychain fvwm m4 mutt fetchmail procmail vim gcal apache2 php5 libapache2-mod-php5 imagemagick mplayer ntp mkisofs aumix wodim cdrecord synergy apt-file sun-java6-jre sun-java6-plugin qiv unsort randomize-lines bittornado flac dbus-x11 subversion mpg321 wmbattery ubuntu-restricted-extras gnucash envyng-core figlet build-essential spamassassin audacity csound nfs-common enscript paps tree texlive ispell lame console-data grip openoffice.org gnugo qgo sanduhr manpages-dev abook remind eclipse git-core duplicity sshfs encfs exim4 xscreensaver xscreensaver-data-extra xscreensaver-gl-extra bsdgames nethack-console
sudo dpkg --force-architecture -i AdobeReader_enu-8.1.3-1.i386.deb
Not too many strange issues except
If one uses the default share name for the virtual box share, it leads to "/sbin/mount.vboxsf: mounting failed with the error: Protocol error". The way to get around this is simply choose a different name for the share than the one than virtual box chooses.
In order to get USB external hard drives to connect up to the virtual machine when they are formatted ext2, ext3, or ext4, I needed to install the fuse-ext2 driver in Snow Leopard (OS X 10.6.1) in order that I didn't get an "unknown drive format" error from OS X. While the drive was plugged in, but virtual box was not running, I was able to specify that the specific drive should pass all the way through to the virtual machine.
Media was playing a bit slow. I found a webpage that said I should simply stop the virtual machine, change the audio settings, change them back, and restart the virtual machine. I did that, and lo and behold it worked.
Google keeps thinking that I want my results in Dutch (Flemish) just because I'm in Belgium. Unfortunately, I've had a rough time figuring out how to get it back into English. I just figured out that there's a "Google in English" link on the Google.com page that doesn't appear if you just use the web browser to submit searches to google. Yay!
Today I discovered that my work computer, reblochon, and my personal computer, chévre, had both been stolen on 2009-10-15. Unfortunately, though I kept meticulous backups of each on the other and both onto an external USB hard disk, the hard disk was stolen, too. It was a very unusual circumstance that both computers and the hard disk were in the same building when I was not, but that is when the tragedy struck. I have no other backups elsewhere, and am starting over pretty much from zero. Kind of intense.
For whatever reason, skype just stopped doing anything. I would type skype and nothing would happen. Eventually, I removed the files shared.lck and shared.xml from ~/.Skype and things started working again. No idea why I thought to do that, or why it fixed things. Just kind of glad it did. Meh.
cat /var/log/dpkg.log | grep "\ install\ " | less
On my birthday (September 11), I decided to take one of my occasional forays into a window manager other than fvwm. This time, I opted to try out the awesome tiling window manager (heretofor referred to as awesomewm), which is really cool and nice, but I ended up dropping for two reasons.
The first reason is that the configuration of the newest version is far from clear (and I don't want to configure the old version while running ubuntu jaunty jackelope only to have to learn the new configuration when karmic koala comes out). I don't mind difficult configuration as long as the documentation is readable and clear, but I didn't feel like learning all of Lua as well as the specific of how it applies to awesome. I just didn't feel like there were answers to the questions I had about the configuration to be found in a well maintained location. In the end, I didn't feel that the fight against the configuration was worth the outcome, especially since...
The second reason was that in the end, I didn't feel comfortable when all of the windows took up all of the space all of the time. I like white space and I think that it helps me to encapsulate thoughts, ideas, and tasks. I found myself trying to adapt by having extra windows open which I wasn't using.
In the end I decided to go back to fvwm, but to return to urxvt (rxvt-unicode) as the main terminal and to add in a few of the features of awesomewm that I liked to fvwm through configuration.
It turns out that rxvt-unicode in my particular environment has problems displaying characters with the correct width. I don't know about other situations, and the urxvt maintainers seem to always claim that any width problems are problems with the system's locale settings. I don't know if this latter statement is true, but what I do know is that applying the following patch found on this bug report page to the source code makes rxvt-unicode display on my systems correctly. It's hard for me to tell where the real problem is, because this patch was rejected by the rxvt-unicode creators.
--- src/rxvtfont.C.bukind 2007-11-30 14:36:33.000000000 +0600 +++ src/rxvtfont.C 2007-11-30 14:39:29.000000000 +0600 @@ -1171,12 +1171,21 @@ XGlyphInfo g; XftTextExtents16 (disp, f, &ch, 1, &g); +/* + * bukind: don't use g.width as a width of a character! + * instead use g.xOff, see e.g.: http://keithp.com/~keithp/render/Xft.tutorial + g.width -= g.x; int wcw = WCWIDTH (ch); if (wcw > 0) g.width = (g.width + wcw - 1) / wcw; if (width < g.width ) width = g.width; + */ + int wcw = WCWIDTH (ch); + if (wcw > 1) g.xOff = g.xOff / wcw; + if (width < g.xOff) width = g.xOff; + if (height < g.height ) height = g.height; if (glheight < g.height - g.y) glheight = g.height - g.y; }
The basic procedure to download, patch, and install the patched version is as follows.
apt-get source rxvt-unicode sudo apt-get build-dep rxvt-unicode
Now edit or patch the file (src/rxvtfont.C).
apt-get source --compile rxvt-unicode sudo dpkg -i rxvt-unicode-9.06_1-.deb
And yay, I now have a well-behaved urxvt with transparency.
The current setup is that I have reblochon (my MacBook Pro running OS X) using VirtualBox to host maroilles (my virtual linux box running Ubuntu 9.04). This generally works well, although maroilles does seem to abort sometimes, and currently this seems to be linked to watching video, but I'm not sure.
The most annoying problem so far is that the key combination Ctrl + arrowKeys seems to stop working sometimes. I think that I'm hitting some combination of Alt, Shift, Cmd, Ctrl, and/or CapsLock to make it happen. I currently think that to get out of the accidental mode, I need to hit and hold Cmd, then hit and release Ctrl, then release Cmd. More updates if necessary.
2010-02-03 EDIT: I found the problem. It seems that there is a missing num-lock on the keyboard, and it's getting pressed. If I have the external keyboard plugged in, it has a key with a symbol on it that looks like a horizontal box with an "X" across it. That key fixes the problem. There must be some other key combination that mimics that key.
2010-09-08 EDIT: I found a solution! The program numlockx (in the numlockx package) allows the numlock to be turned on and off from the commandline. After a quick install, the command
numlock toggle
fixes the problem. Yay! (Just for note, it seems that it's the X windows numlock that's getting pressed, not the actual numlock on the keyboard.)
For some inexplicable reason, I was getting a "bash: Permission denied" error on running a script that I was writing. I looked at all of the permissions on the file and tried to figure everything out. Eventually I realized that I was running the script in a directory mounted from a USB hard drive. I took a look at the mount options and sure enough, the drive was mounted "noexec". I added "exec" to the options in /etc/fstab and remounted the USB drive. Now the script works with no problem.
Tags: Duh!
This is just a reminder to myself of how to use ImageMagick to take pictures and stick 'em together.
montage -mode Concatenate image1.jpg image2.jpg image3.jpg final.jpg
Well, upon installing a new 500 G, 5400 rpm harddrive into goat, I decided to upgrade the Ubuntu as well since I would need to install on the new drive anyway. The hardware and OS install went incredibly smoothly. I took the old 120 G drive and put it in a USB enclosure. I used it immediately to copy my home directory back into place, and it now stands as a backup of goat.
Since I'm in Belgium and trying to learn French now, I decided to change "goat" to "chèvre". I wonder what issues will crop up with using a UTF8 character in the hostname. Perhaps none, but I doubt that.
I ran into a weird problem where the machine wouldn't shutdown properly. When the wireless card went to turn off, I got a deluge of errors starting with.
iwlagn: Error: Response NULL in 'REPLY_ADD_STA'
I googled that and found a solution of installing the linux-backports-modules-jaunty package, and this fixed the problem.
I also found that I like wicd much better than the Gnome wireless controls although I though I saw a Gnome wireless manager on a different Jaunty install that was much nicer that what I found in this install. Whatever. wicd it is.
I also needed some special access to the mixer and microphone controls and I couldn't figure out which of the Gnome programs I had used in order to manipulate that before. It turns out it was gnome-volume-control.
(Note: I'm continuing notes in this post because they effectively become a list of things to do on installing ubuntu on a D830.) 2009-08-12 - I found a page detailing what I need to do to get the synaptics touchpad controls working better. That page is https://help.ubuntu.com/community/SynapticsTouchpad, but here are the end results: Create a file, /etc/hal/fdi/policy/shmconfig.fdi, and add the following to it.
<?xml version="1.0" encoding="UTF-8"?> <deviceinfo version="0.2"> <device> <match key="input.x11_driver" string="synaptics"> <merge key="input.x11_options.SHMConfig" type="string">True</merge> </match> </device> </deviceinfo>
On reboot, SHMConfig was available to programs like synclient and gsynaptics. I use the following in a startup script for me.
/usr/bin/synclient AccelFactor=0.2 MaxTapTime=0 HorizEdgeScroll=1 &
I will note here that I did try to use "chèvre" (note the accent aigu) as my hostname. It seemed to be working fine, until I did a fetchmail. It looks like SMTP (or at least the programs running SMTP that I deal with) don't like non-ASCII characters. For now, the hostname is now "chevre", but I use the accented name in my prompt and in the welcome message.
It took a while to figure out how to add and remove a permanent DNS entry with NetworkManager, but I figured it out after awhile. This was necessary because I was getting crappy DNS service from a couple of the routers I was using in Italy.
The file to change is /etc/NetworkManager/system-connections/Auto\ eth0 (or whichever interface you want to affect). To add the DNS entry include the following lines.
dns=208.67.222.222 ignore-auto-dns=true
In order to reset, just remove the "dns=" line and set the other value to "false".
I bought a 500GB external USB/Firewire disk for backups. I intend to use it to back up my home and media directories. I had forgotten a couple of steps in the process of formatting and mounting an external USB drive, so I'm writing them down again here.
sudo fdisk /dev/sdb
I then chose to "d" delete the one partition already on the drive. Then "n" to create a new partition and chose primary partition 1. Since this is an external drive, I used the entire drive as one partition. I then chose "w" to write to disk and exit.
sudo mkfs.ext3 -m1 /dev/sdb1 sudo tune2fs -O^dir_index /dev/sdb1
I changed this filesystem option, because this drive will be mounted on an OS X MacBook Pro sometimes and apparently the ext2/ext3 filesystem mounter program available for OS X can't write to an ext2/ext3 filesystem which has dir_index set (the default).
cd ~ mkdir SAUVEGARDE
This is the place where I want the drive to be "automatically" mounted. I edited /etc/fstab to include the line
/dev/sdb1 /home/nodog/SAUVEGARDE auto rw,user,noauto 0 0
mount SAUVEGARDE sudo chown nodog:nodog SAUVEGARDE
This previous command was one that I'd forgotten, so I was stuck with root owning ~/SAUVEGARDE once it was mounted until I remembered.
nice rsync -avP PERILO SAUVEGARDE
Now I just need to fix my backup program so that it will work with SAUVEGARDE as a destination/source instead of just a machinename.
I've been working in the Mac OS X terminal quite a bit lately, and screen (a virtual terminal program, which is one of my favorite programs of all time) has been having trouble wrapping at the end of the line at the bottom of the terminal. It's really been driving me a bit bonkers, but I finally found the solution, though there is still some testing to do.
In Terminal's preferences under Advanced, one can choose the terminal type that Terminal declares. I switched this from xterm-color to vt102, and my line-wrapping/disappearing problems went away. I was worried that I would lose directory colors, but they seem to still be working. I see that there may be reasons to choose one of the other terminal types to emulate, but vt102 is working for now. Yay!
Holy crap, this is cool! I never knew before that ssh can do reverse port forwarding. My work machine is on a local private network, and I don't have administrator privileges on the router, but I want to be able to connect to it from my home machine. Fortunately, I do have administor privileges at home, so I've already set up port forwarding for ssh.
On the local machine (work machine, here), one runs something like the following before physically leaving the local machine.
ssh -R 8022:localhost:22 myhome.machine.net
Upon arriving home, one connects via ssh to the then-remote-now-local host via a new local port number.
ssh -p 8022 localhost
And voila, one is connected back. Yay! I wrote a small script to rerun the first command if it ever goes down for any reason (like I reboot my machine at home, or the IP address of my home machine changes). (And, yes any actual port numbers have been changed to protect the innocent.) All of this was gathered from a great ssh tutorial.
Well, I gave up on dual booting reblochon and then accessing the linux partition as a virtual machine. It turns out that accessing the physical partition on a Mac OS X host system is a bit more limited and problematic than I wanted to deal with. I opted to just have a completely virtual machine. It has it's own problems, too, but it does let me access linux in a way that I like at the same time that I am able to access and learn Mac OS X.
So, maroilles is the virtual linux machine inside of the Mac OS X of reblochon. Setting it up has been mostly just following the instructions here and there. Unfortunately (in some sense), I like to use fvwm2 instead of Gnome or KDE, so that makes life a little more difficult when doing the configuration. The most interesting thing that I've run across, though is a new way of configuring FreeType fonts to use in xterm. I like xterm because it includes unicode support. Here's an example of the new syntax. Note the use of the "-fa" instead of the "-fn" option.
xterm -fa "DejaVu Sans Mono:style=Bold:pixelsize=14"
openssh-server keychain fvwm m4 mutt fetchmail procmail vim-full gcal apache2 php5 libapache2-mod-php5 imagemagick mplayer audacious ntp mkisofs aumix k3b wodim cdrecord synergy apt-file sun-java6-jre sun-java6-plugin qiv randomize-lines bittornado flac dbus-x11 subversion mpg321 wmbattery ubuntu-restricted-extras gnucash envyng-core figlet build-essential spamassassin audacity csound nfs-common enscript paps tree texlive ispell lame console-data grip openoffice.org gnugo qgo kontact sanduhr manpages-dev
NOTE:To add next time: abook remind
NOTE:To remove next time: kontact
sudo dpkg --force-architecture -i AdobeReader_enu-8.1.3-1.i386.deb
Work gave me a new MacBook Pro (apparently ver 5.1) and I'm calling it reblochon. I'm trying to dual boot, and also run my linux partition as a virtual machine under Mac OSX using Virtual Box. I'll be documenting some of my process here.
to add a user to a group
dscl . -append /Groups/operator GroupMembership username
Erg. I just spent far too much time chasing down a bug caused by "The Microsoft Exchange POP3 service." Basically, Microsoft's usage of the POP3 protocol is broken (as expected, I guess), so one needs to add "auth ntlm" to the .fetchmailrc "poll" section for a POP3 server with Microsoft Exchange installed. That seemed to fix the problem, but Bleh! Below are some of the error messages I was getting.
Logon failure: unknown user name or bad password. Error exchanging credentials Authorization failure on
Oh, and I found a neat, slightly-not-documented keyword for the .fetchmailrc. Use "skip" instead of "poll" in the .fetchmailrc and a server won't be polled unless specifically named on the commandline. It's a good way to only check a server every once in a while.
Well, I've been aching to play some Pac-Man since I've moved to Brussels, and since I have all of those MAME ROMs sitting around, I decided to do something about it. sdlmame is a really good answer to the MAME question these days. I understand that there's even a really good menu program that works well with it, but I can't remember the name of that at the moment.
Unfortunately, I was getting some nasty terrible sound quality after about 1 minute of gameplay. It was like suddenly the volume went to maximum 1bit sound. Bleh. I remembered that I had changed the default sample rate from 48000 to 22050 in order to fix some problems with the sound of Pac-Man. I changed this to 44100, and I haven't heard the problem again. Awesome!
I didn't realize that I even had an internal microphone until a friend with a D830 said she used her microphone. I did some web searching, and I found this list of steps to make it happen in Gnome. I changed to a Gnome session and tried to follow these instructions, though I had to fumble around a little to figure out what to do.
- choose volume control - choose preferences - turn capture off and input source off - turn the recording tab volume up - set the mic to not be muted - under options, choose the front mic
The correct settings for sound in Skype were as follows
in: HDA Intel( hw:Intel,0) out: pulse ring: pulse
The service providers at the Cité Universitaire (Magellen) told me that I need to use an http proxy. I did so by adding the line
export http_proxy=http://squid.magellan.fpms.ac.be:3128
to the appropriate section of my .bashrc.
I'm trying to install ING's Home'Bank and I got this error.
libtiff.so.3 : cannot open shared object file: No such file or directory
Creating a link to libtiff.so.4 seems to be the answer.
cd /usr/lib (or /usr/lib32 for a 64 bit system) sudo ln -s libtiff.so.4 libtiff.so.3
Then
libexpat.so.0 : cannot open shared object file: No such file or directory
but I couldn't follow their suggestion of installing the compat-expat1 package. I did a similar thing to the above.
cd /lib sudo ln -s libexpat.so.1 libexpat.so.0
And again
libstdc++-libc6.2-2.so.3 : cannot open shared object file: No such file or directory
This one seems trickier. I downloaded the package libstdc++2.10-glibc2.2_2.95.4-24_i386.deb from href="http://packages.ubuntu.com/dapper/i386/libstdc++2.10-glibc2.2/download found by searching for "libstdc++-libc6.2-2.so.3 ubuntu" in google. Then
sudo dpkg -i libstdc++2.10-glibc2.2_2.95.4-24_i386.deb
And it finally works!
EDIT 2009-11-15: It turns out that there is a package called homebank in the repository and ING's package will try to upgrade to that package on an update. To stop this behavior, one must "hold" the homebank package.
sudo aptitude hold homebank
I was at a friend's house today and none of my mail would go out. It turns out that his ISP, grandenetworks.net, was blocking port 25. Well, I went to the file /etc/exim4/conf.d/transport/30_exim4-config_remote_smtp_smarthost and added the line
port = 587
and then ran
sudo update-exim.conf sudo /etc/init.d/exim4 reload
and nothing was fixed. After reading on the internet for a while, I realized that I'd chosen "dc_use_split_config='false'" in my /etc/exim4/update-exim4.conf.conf file. This means that I need to add the "port = 587" line to the transport/30_exim4-config_remote_smtp_smarthost section of /etc/exim4/exim4.conf.template. Once I did that and ran the two commands listed above, everything seemed to work fine. Yay!
Okay, so I'm learning French, and I'm trying to figure out how to type the French characters. I've finally decided that I'll be happy enough using the compose key to make the characters. I like the idea of understanding what characters that I'm putting together and not having to memorize new weird placements for them. If I were going to go all out, I might learn a french dvorak keyboard, but I think I'll stick to the compose key for a while. This negates all of the work that I did before and before on the Esperanto layoout.
Here is the contents of the current ~/bin/asdf
#!/bin/bash setxkbmap us dvorak -option ctrl:nocaps -option compose:caps
The "ctrl:nocaps" option effectively removes the default behavior of capslock, and "compose:caps" makes capslock the compose key. It took me a long while to figure out that combination. Here's the contents of ~/bin/aoeu
#!/bin/bash setxkbmap us -option
The empty option removes any previous options from previously loaded keyboards. I guess I will see how I like the compose key. (I seem to remember not liking it before.
àèìòù áéóíú âêîôû äëïöü ç œ «» ĉĝĥĵŝŭ
NOTE: At some point in this process, my xkb setup got severely munged and I was getting errors about loading the layouts. I solved this by reinstalling the packages libxkbfile1, x11-xkb-utils, and xkb-data.
Ripping the audio off DVDs into WAV files.
mplayer -vc null -vo null \ -ao pcm:fast:waveheader:file=audioFilename.wav \ dvd://1 -chapter 8-8 -af resample=44100 -aid 130
Since upgrading to ubuntu 8.10 (intrepid), my touchpad has been much sadder than before. Here's what I did to get it back into shape. The big deal is that now the X11 input stuff is all handled by hal, which is great for hotplugging, but somewhat a pain for some other reasons.
First, following the instructions at https://help.ubuntu.com/community/SynapticsTouchpad, I put the following lines into /etc/hal/fdi/policy/shmconfig.fdi and then rebooted.
<?xml version="1.0" encoding="ISO-8859-1"?> <deviceinfo version="0.2"> <device> <match key="input.x11_driver" string="synaptics"> <merge key="input.x11_options.SHMConfig" type="string">True</merge> </matc> </devic> </deviceinf>
Having done that, I could now run synclient AccelFactor=0.2 (which I put into my myfakefvwmpanel startup script) and it changes the touchpad acceleration. Yay! There are a lot more parameters that can be tweaked using synclient, but the better way to edit them is using gsynaptics, and then look at the parameters afterwards with synclient -l.
So, I've been trying to get ChucK running on goat for a while. The first problem has been that ChucK is not 64-bit clean, so compiling it on goat wasn't working. Things would look fine, and then I'd get no sound. I finally solved this problem by just compiling ChucK on a 32-bit system and then copying the executable over to goat. It worked, although sometimes Firefox's audio control gets in the way. They're unhappy with each other, but that's a small price to pay to have ChucK on my fast machine.
The next problem that I ran into was the HID devices weren't available because of permissions problems. The solution is to add a group (hidaccess) to /etc/group and to edit /etc/udev/rules.d/20-names.rules to include the following lines.
# Input devices, group under /dev/input KERNEL=="event[0-9]*", NAME="input/%k", GROUP="hidaccess" KERNEL=="mice", NAME="input/%k" KERNEL=="mouse[0-9]*", NAME="input/%k", GROUP="hidaccess" KERNEL=="js[0-9]*", NAME="input/%k", GROUP="hidaccess" KERNEL=="ts[0-9]*", NAME="input/%k" KERNEL=="uinput", NAME="input/%k"
I have no idea why, but for some reason with Ubuntu 8.10 Intrepid Ibex the way my Dvorak Esperanto xmodmap keyboard layout. The way that I was doing it was this.
With Intrepid Ibex, it seemed that gnome-settings-daemon was returning control, but then working in the background and taking long enough that the effects of asdfeo were being overwritten.
I decided that it was time to bite the bullet and create an xkb map so it could be loaded by gnome-settings-daemon. The map file, dvorak-eo.xkb, was relatively easy to create. I inserted it into the file /usr/share/X11/xkb/symbols/us. At this point, I can type the command
setxkbmap "us(dvorak-eo)"
and it is loaded and gnome-settings-daemon seems to remember that is the map I want when starting in X the next time. Yay!
The one thing I couldn't figure out how to do was set up the files /usr/share/X11/xkb/rules/base.xml and /usr/share/X11/xkb/rules/base.lst so that gnome-keyboard-properties can show it in the menu. I followed the instructions at the following page, but my layout never showed up as an available choice, even after restarting X or rebooting the computer.
I even changed some of the names in those files, just to see if I could affect the menus presented by gnome-keyboard-properties, but no change at all was ever seen.
EDIT: Well, I was wrong about setxkbmap being remembered on reboot. At this point, I'd moved my .gconf directory away, and then let gnome-settings-daemon recreate it, and then moved my .gconf directory back. I have no idea how, but somehow that fixed the "delay" problem described above and my old asdfeo script is working again now. Whatever.
openssh-server keychain fvwm m4 mutt fetchmail procmail vim-full gcal apache2 php5 libapache2-mod-php5 imagemagick mplayer audacious ntp mkisofs aumix k3b wodim cdrecord synergy apt-file sun-java6-jre sun-java6-plugin qiv randomize-lines bittornado flac dbus-x11 subversion mpg321 wmbattery ubuntu-restricted-extras gnucash envyng-core figlet build-essential spamassassin audacity csound nfs-common enscript tree texlive ispell lame console-data grip openoffice.org gnugo qgo kontact
sudo dpkg --force-architecture -i AdobeReader_enu-8.1.3-1.i386.deb
Well, I did the standard upgrade, and there were two major problems. The first was WIFI and the second was the keyboard was doing some very weird things.
The WiFi problem seems to be related to using the WPA and WPA2 with the Intel Pro 4695 AGN driver, iwlagn. I switched back to WEP for now at home, but I imagine that there will be problems if I try to use a secure WPA or WPA2. There is something called wpa_gui that seems to be oriented toward the wpa_supplicant protocols, but I don't know how yet.
The keyboard was doing all kinds of wierd things. The arrow keys wouldn't work sometimes, then only one of the Control key would work, then the Page Up and Page Down keys would stop working only in Firefox. I'm not sure how the keyboard problems started, but they seem to have been fixed by going back through
sudo dpkg-reconfigure console-setup
I'm now left with two niggling problems. One is that the touchpad seems to be controlled by "HAL" instead of xorg.conf, and I'm not sure how to make all the acceleration settings for it. The other problem is that my Esperanto Dvorak script won't run from my .fvwm2rc file. I guess I'll figure all this out over the next couple of days. Oh, yeah, and I'll need to figure out the WPA supplicant stuff.
EDIT:I went back and did a re-install. It works MUCH better now.
This actually went pretty smoothly. It didn't look pretty, but it did happen pretty smoothly.
Using ubuntu and a linksys router, I couldn't get connectable for my bittorrent clients. I had already directed ports through the router to the machine I would be bittorrenting on, but ubuntu was giving me no love. It turns out that I needed to do two things. The first was I needed to limit the port range that my bittorrent client was using. That wasn't too difficult. Then I needed to use the following command.
iptables -A INPUT -p tcp -d 0/0 -s 0/0 --dport 4940:4953 -j ACCEPT
where "4940:4953" are the start and end numbers of the ports I used (actually values changed to protect the innocent).
Since I switch our WIFI network over to WPA2, the wifi on goat hasn't started on boot up. I looked in the log (/var/log/syslog) and I saw that there was an error about "link is not ready." I looked on the web and found a similar problem in the ubuntu forums and someone added the line
pre-up sleep 20
to /etc/network/interfaces for the wlan0 network interface. On reboot, this config works just fine. Yay!
I've been trying to figure out why goat hasn't been using my .procmailrc and I finally found the answer (in the procmail man page, of course).
If procmail is not installed globally as the default mail delivery agent (ask your system administrator), you have to make sure it is invoked when your mail arrives. In this case your $HOME/.forward (beware, it has to be world readable) file should contain the line below. Be sure to include the single and double quotes, and unless you know your site to be running smrsh (the SendMail Restricted SHell), it must be an absolute path.
"|exec /usr/bin/procmail"
Some mailers (notably exim) do not currently accept the above syntax. In such case use this instead:
|/usr/bin/procmail
goat uses exim as the MTA, so I had to update my .forward file to look like the latter example.
I had to shut down lighthill today. I no longer have my own computer on the UT campus anymore. Oh, well. I took the 250G drive I store my music on and shoved it into pinkmame, so pinkmame is actually working as a server now and not just a MAME machine.
After putting hardy heron on lighthill, all of the problems on goproblems.com gave an error starting with the following.
Invalid SGF data. Sorry, you're screwed. Error message: illegal start of token string:
The solution is to remove the package icedtea-gcjwebplugin and the package sun-java6-plugin becomes the default. It's a problem with Hardy's default java plugin.
Putting hardy heron on lighthill was a snap. From the commandline, I executed
sudo aptitude install update-manager-core sudo do-release-upgrade
and after the process finished, I chose to reboot. Everything looks good except unsurprisingly, the dual-headed-ness didn't come back, so I just followed the same procedure as last time last time with one addition.
sudo cp mga_drv.so mga_drv.so-xorg-1.4.0.90-backup sudo cp mga_drv.so-from_matrox-4.4.3-7.2.0-backup mga_drv.so sudo /etc/init.d/gdm restart
Yay!
When trying to run audacity on hardy heron on lighthill, I was getting lots of errors and it wouldn't start. Here's an example of the error messages.
lighthill:~/PERILO/music/$ audacity 02-2,1.mp3 jackd 0.109.2 Copyright 2001-2005 Paul Davis and others. jackd comes with ABSOLUTELY NO WARRANTY This is free software, and you are welcome to redistribute it under certain conditions; see the file COPYING for details JACK compiled with System V SHM support. loading driver .. apparent rate = 48000 creating alsa driver ... hw:0,0|hw:0,0|1024|2|48000|0|0|nomon|swmeter|-|32bit control device hw:0 configuring for 48000Hz, period = 1024 frames (21.3 ms), buffer = 2 periods ALSA: final selected sample format for capture: 16bit little-endian ALSA: use 2 periods for capture ALSA: final selected sample format for playback: 16bit little-endian ALSA: use 2 periods for playback Expression 'ret' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 1034 Expression 'AlsaOpen( hostApi, parameters, streamDir, &pcm )' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 1066 Expression 'ret' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 1034 Expression 'AlsaOpen( hostApi, parameters, streamDir, &pcm )' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 1066
The last two errors would just repeat over and over.
It turns out that the jackd program was the problem. I did a
sudo aptitude remove jackd
and audacity had no problems. It took some searching on the forums to find the answer.
For reasons that I won't go into, I needed to load MATLAB R14 SP2 running on 64-bit ubuntu Linux 8.04. I installed from the distribution CDs just fine (some people seem to have trouble with this), but was getting nasty looking errors about corrupted memory when I tried to run MATLAB.
*** glibc detected *** /usr/local/matlab/bin/glnxa64/MATLAB: malloc(): memory corruption: 0x0000000000596b90 *** ======= Backtrace: ========= /lib/libc.so.6[0x7f26dec62a14] /lib/libc.so.6(__libc_calloc+0x10f)[0x7f26dec6407f] /lib64/ld-linux-x86-64.so.2(_dl_allocate_tls+0x3c)[0x7f26e063532c] /lib/libpthread.so.0(pthread_create+0x558)[0x7f26df4bfb88] /usr/local/matlab/bin/glnxa64/libmwservices.so[0x7f26e026dbd7] /usr/local/matlab/bin/glnxa64/libmwservices.so[0x7f26e026d9de] /usr/local/matlab/bin/glnxa64/libmwservices.so[0x7f26e02429f7] ...
The fix to this so far has been to recognize the heap problems, but let MATLAB keep running anyway. To do this, set
export MALLOC_CHECK_=1
or just run MATLAB with
MALLOC_CHECK_=1 matlab
I don't know of any ill effects of having this be globally on, so I'm just going to add it to .bashrc.
EDIT: Keeping that variable set all the time causes lots of debugging messages to occur all the time, so I'll just turn it on for matlab when I need to.
Some programs, like evince, were taking a long time to start up, giving dbus erros. I solved this by adding the package dbus-x11. Yay!
EDIT: Well, this didn't work completely. I'm still getting the following errors from evince on startup.
** (evince:26069): WARNING **: Service registration failed. ** (evince:26069): WARNING **: Failed to connect to socket /tmp/dbus-IWKKLDHkVF: Connection refused
It turns out that the 75dbus_dbus-launch script in /etc/X11/Xsession.d is not installed from the dbus-x11 package for ubuntu 8.04. This was done because ubuntu assumes that people are going to be using Gnome, KDE, or XFCE, and that script causes problems with those environments. I copied that script over from lighthill to goat.
RE-EDIT: Okay, so I've now removed /etc/X11/Xsession.d/75dbus_dbus-launch in lieu of a better way of doing things. I've added the following lines to my program myfakefvwmpanel which is called by .fvwm2rc.
if which dbus-launch >/dev/null && test -z "$DBUS_SESSION_BUS_ADDRESS"; then eval `dbus-launch --sh-syntax --exit-with-session` fi
I've also added the following lines to myfakefvwmpanel in order to get Gnome and KDE applications running properly. It does take up memory, but it does make them run faster, too.
# Make GTK apps look and behave how they were set up in the gnome config tools if which gnome-settings-daemon >/dev/null; then gnome-settings-daemon & fi # Preload stuff for KDE apps if which start_kdeinit >/dev/null; then LD_BIND_NOW=true start_kdeinit --new-startup +kcminit_startup & fi
When I boot up on wifi, the resolv.conf file gets munged. I suspect it's DHCP doing it, even though I don't have any DHCP turned on in /etc/network/interfaces. I found some notes in a forum that suggest I should add the following before the "request" line in /etc/dhcp3/dhclient.conf.
prepend domain-name-servers 123.45.678.90, 123.45.678.91;
with adjustments made for my own nameservers. I'm trying it.
EDIT: It seems to have worked.
For some reason, I can't get postfix working on goat through dreamhost. I followed my own instructions from 2006-10-12 which are basically the same as dreamhost's instructions for postfix, but even though the mail log said it went through, I never received the emails.
So, I decided to switch to exim4 and see if that worked and following dreamhost's instructions for exim, it worked great. Basically, all that I needed to do was put my mailhost:username:password into /etc/exim4/passwd.client and then do a
dpkg-reconfigure exim4-config
so that my config file, /etc/exim4/update-exim4.conf.conf, looked like
dc_eximconfig_configtype='smarthost' dc_other_hostnames='' dc_local_interfaces='127.0.0.1' dc_readhost='konfuzo.net' dc_relay_domains='' dc_minimaldns='false' dc_relay_nets='' dc_smarthost='mail.konfuzo.net' CFILEMODE='644' dc_use_split_config='false' dc_hide_mailname='true' dc_mailname_in_oh='true' dc_localdelivery='mail_spool'
I also added the line
port = 587
to the file /etc/exim4/conf.d/transport/30_exim4-config_remote_smtp_smarthost in order to switch from port 25 to 587 for smtp. That also worked without a hitch. Yay!
The monitor on bleu died suddenly, so I decided to get a new laptop. I chose a Dell Latitude D830 because I know it's compatible with Linux and is available with an educational discount from UT. I will call the new laptop goat.
As I was transferring the files from bleu to sedusa for backup, the hard drive in bleu died as well. Luckily, I got everything that was unique on bleu onto sedusa before the crash. Yow!
In order to specify the refresh rate in xorg.conf, one uses the "Modeline" entry. To do this for a specific monitor, one puts the Modeline entry in the "Monitor" section and then references that Modeline's title in the "Screen" section. In order to get a Modeline entry for a specific resolution and refresh rate, one uses the gtf command, for example
gtf 1280 1024 75
Here are the resulting sections in xorg.conf.
Section "Monitor" Identifier "Hitachi CML175SXW" HorizSync 24.0 - 80.0 VertRefresh 56.0 - 75.0 # 1280x1024 @ 75.00 Hz (GTF) hsync: 80.17 kHz; pclk: 138.54 MHz Modeline "1280x1024_75.00" 138.54 1280 1368 1504 1728 1024 1025 1028 1069 -HSync +Vsync EndSection Section "Screen" Identifier "Primary Screen" Device "nvidia Quadro2 Pro 0" Monitor "Hitachi CML175SXW" DefaultDepth 24 SubSection "Display" Depth 24 Modes "1280x1024_75.00" EndSubSection EndSection
I don't know why, but audacious and audacity were both spouting errors and behaving badly on my ubuntu gutsy 7.10 installation on lighthill. Both of these were recently upgraded in the repositories, so I wanted to go back to the older versions, so at least they'd work. Here was the full procedure necessary to do that.
sudo aptitude install audacity=1.3.3-1ubuntu0.1 sudo aptitude install audacious=1.3.2-4 sudo aptitude hold audacity audacious audacious-plugins audacious-plugins-extra
I had to use apt-cache showpkg audacious | less and apt-cache showpkg audacious | less to find the version numbers. Now audacious audacity and aumix are all held back during upgrades (must be something about the "au" at the front of the program name).
from ubuntuguide.org:
sudo /usr/share/doc/libdvdread3/install-css.sh
Well, the dpkg --set-selections command line that I'd used before didn't seem to be working on usagi for some reason. I found an equivalent line in aptitude so here is the complete new set of commands to get aumix properly installed on gutsy.
sudo apt-get build-dep aumix sudo apt-get source --compile aumix sudo dpkg -i aumix_2.8-18_i386.deb sudo aptitude hold aumix
Here is a good tutorial on how to turn regular video files into a DVD using tovid. The basic steps are
todisc -files File1.mpg File2.mpg File3.mpg \ -titles "Episode 1" "Episode 2" "Episode 3" \ -out Season_one makexml -menu Menu.mpg foo1.mpg foo2.mpg foo3.mpg -out MyDisc makedvd -burn MyDisc.xml
In order to bring the quality up to 10 (instead of the default of 6), I ended up using the following command line with tovid beforehand.
tovid -quality 10 -ntsc -dvd -in foo.avi -out foo-encoded
Here is another really good tutorial!!!
(EDIT) Actually, I had a problem with the results of the tovid command above. The input and output movie files were jumping all over the place in mplayer, so using mplayer to encode it didn't make much sense. I ended up using a ffmpeg command line which worked great.
ffmpeg -i myfile.avi -target ntsc-dvd dvd.mpg
There are a couple of programs designed to wipe files on a filesystem, shred and wipe. It turns out that journaling filesystems don't really allow "in place" rewriting of a file, so these programs are pretty useless for wiping files. However, an entire partition or disk can be wiped at once by specifying the device as the file to be wiped. Using this technique, dd can also be used. In the following three examples "/dev/hda" should be replaced with the drive or partition (example: /dev/hdb3) you want to erase. Note that 1 pass on a USB connected hard drive looks like it's going to take about 22 hours!
shred -v /dev/hda wipe -kD /dev/hda dd if=/dev/urandom of=/dev/hda
Realistically, just removing the partition information from the drive probably good enough for keeping 99.99% of all people from recovering the data. Beyond that, writing random data or zeros once probably covers 99.99% of all situations beyond that. However, writing the data 25 times with the specified data patterns is considered 100% safe.
I was driving myself a little batty, trying to figure out why I couldn't print to another cups printer using the same method as last time. Eventually, I figured out that the "Manage Server -> Share published printers connected to this system" option wasn't checked on the machine the printer was actually connected to. I think that this is the second time I've learned this lesson.
Wow, the subversion import command is easy to misunderstand. One imports from a directory that one won't necessarily be using, but (at least for me) imports into the directory one will be using, then goes into the directory above the new directory and does the initial checkout there. Yeeks.
Example:
cd ~/phyd/dissertation mkdir aim-tmp touch aim-tmp/goals.uxt svn import aim-tmp file:///home/nodog/repository/phyd/dissertation/aim svn checkout file:///home/nodog/repository/phyd/dissertation/aim rm -r aim-tmp
We were running out of space on the main hard drive of sedusa. It had a 20G system partition and a 35G home directory partition and that just wasn't enough. So I'm replacing it with an 80G drive which will be split 15G/63G (a 2G swap partition).
I put the new drive on the system in place of the CDROM which had been the master on the second IDE controller.
I used fdisk /dev/hdc and created three partitions on the new drive: 15G, 2G, and the rest. I used mkfs.ext3 -m 1 /dev/hdc1 and mkfs.ext3 -m1 /dev/hdc3 to create the filesystems on the two filesystem partitions and mkswap /dev/hdc2 to format the swap partition. I also set /dev/hda1 as bootable.
I booted the system into single user mode, umounted the /home directory (after first umounting /home/nodog/PERILO), mounted the new system partition with mount /dev/hdc1 /mnt, and copied the filesystem with cp -prvx /* /mnt. This stopped working somewhere in /proc so I had to restart it, only copying the directories beyond /proc. I think this will be no problem, because I understand that /proc is recreated on boot.
I umounted /mnt and mounted the new home directory partition on /mnt with mount /dev/hdc3 /mnt, remounted /home with mount /home, and then copied the home directories with cd /home; cp -prvx * /mnt.
NOTE: I probably could've avoided booting from CD by chroot-ing here, editing /etc/fstab, editing /boot/grub/menu.lst, and running update-grub, but I didn't think of it as things were happening.
At this point, I connected the new drive in the same location that the old drive had been (master on the primary IDE controller) and reconnected the CDROM drive.
I booted from the ubuntu alternate CD and went into rescue mode. I started a shell with /dev/hda1 as the root. I edited /etc/fstab to reflect all the changes I'd made, and then edited /boot/grub/menu.lst so that (hd0,0) was the boot drive and the "root=" option pointed to "/dev/hda1". I did all this in the area of the file that lets update-grub automatically handle the changes.
For some reason, there was a problem with the X resolution. I'm not sure why. It was fixed pretty easily with dpkg-reconfigure xserver-xorg.
Admittedly, things didn't go quite this smoothly, but I think that they could've if I'd done things right the first time.
So, the hard drive crashed on usagi, and luckily it happened only after I'd getten everything in ATTIC backed up to another partition on another drive. Unfortunately, that did mean that I lost my dualheaded xorg.conf, my apache config, and my aumix installation. The last two weren't too hard to recreate, but the xorg.conf file along with the downgraded ATI driver were a bit harder to recreate. After doing the work, here's the xorg.conf for usagi.
Section "ServerFlags" Option "xinerama" "true" Option "DefaultServerLayout" "DualHead" EndSection Section "ServerLayout" Identifier "DualHead" Screen "Primary Screen" 0 0 Screen "Secondary Screen" RightOf "Primary Screen" InputDevice "Generic Keyboard" InputDevice "Configured Mouse" EndSection Section "Files" FontPath "/usr/share/X11/fonts/misc" FontPath "/usr/share/X11/fonts/100dpi/:unscaled" FontPath "/usr/share/X11/fonts/75dpi/:unscaled" FontPath "/usr/share/X11/fonts/Type1" FontPath "/usr/share/X11/fonts/100dpi" FontPath "/usr/share/X11/fonts/75dpi" FontPath "/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType" EndSection Section "Module" Load "i2c" Load "bitmap" Load "ddc" Load "extmod" Load "freetype" Load "glx" Load "int10" Load "type1" Load "vbe" EndSection Section "InputDevice" Identifier "Generic Keyboard" Driver "kbd" Option "CoreKeyboard" Option "XkbRules" "xorg" Option "XkbModel" "pc104" Option "XkbLayout" "us" EndSection Section "InputDevice" Identifier "Configured Mouse" Driver "mouse" Option "CorePointer" Option "Device" "/dev/input/mice" Option "Protocol" "ExplorerPS/2" Option "ZAxisMapping" "4 5" Option "Emulate3Buttons" "true" EndSection Section "Monitor" Identifier "Hitachi CML175SXW" HorizSync 24.0 - 80.0 VertRefresh 56.0 - 75.0 EndSection Section "Monitor" Identifier "ViewSonic VG900b" HorizSync 30.0 - 82.0 VertRefresh 50.0 - 75.0 EndSection Section "Device" Identifier "nvidia Quadro2 Pro 0" Driver "nv" BusID "AGP:1:0:0" EndSection Section "Device" Identifier "ATI Radeon 7500 RV200 QX" Driver "radeon" BusID "PCI:4:13:0" EndSection Section "Screen" Identifier "Primary Screen" Device "nvidia Quadro2 Pro 0" Monitor "Hitachi CML175SXW" DefaultDepth 24 SubSection "Display" Depth 24 Modes "1280x1024" EndSubSection EndSection Section "Screen" Identifier "Secondary Screen" Device "ATI Radeon 7500 RV200 QX" Monitor "ViewSonic VG900b" DefaultDepth 24 SubSection "Display" Depth 24 Modes "1280x1024" EndSubSection EndSection
openssh-server keychain fvwm m4 mutt fetchmail procmail vim-full gcal apache2 php5 libapache2-mod-php5 imagemagick mplayer audacious ntp mkisofs aumix k3b wodim cdrecord synergy apt-file sun-java6-jre sun-java6-plugin qiv randomize-lines bittornado flac dbus-x11
In the wake of oink's demise, google's advanced search features have been most helpful. Here's an example of some great searches:
intitle:index.of +mp3 +"advance" -html -htm -php -txt -pl
To search for specific titles, replace "advance" with what you want.
To search for software, replace "mp3" with "zip" or "iso".
To search for full albums, replace "mp3" with "rar".
Here's an example of it as a URL: an example of Google linking advance releases.
I started using audacious as a music player instead of xmms just recently. I'm not sure why I got dissatisfied with xmms, but I think it had something to do with the weird times listed for newer mp3s and other file formats. In searching that, I found that xmms wasn't under development anymore, so I decided to switch to it's apparently current successor, audacious.
Two problems seemed to pop up with this. There's apparently a situation when moving the audacious window without using an fvwm title bar that causes a temporary locking of fvwm. According to Audacious' FAQ this problem is solverd by adding the line
DestroyFunc EWMHActivateWindowFuncto your .fvwm2rc file.
The second problem was that I couldn't seem to get rid of the title bar using fvwm's Styles. The main problem was that I wasn't able to correctly specify the name or class of audacious' windows. Finally, I found the program xprop which works similar to xwininfo, but gives the window's class name as well. I'd chosen "audacious" for the style's window name. It turns out that simply changing to "Audacious" solved the problem because this is the class name for audacious' windows (window names in Style lines are case-sensitive). Here's the style line.
Style "Audacious" !Title, BorderWidth 0, ClickToFocus
I was getting an error ("aumix: SOUND_MIXER_READ_DEVMASK") on gutsy when I ran aumix. This is a known gutsy bug. Someone suggested that they got it running from source. I'd never done that before, so I needed to figure out how, and then do it. Here are the steps.
sudo apt-get build-dep aumix sudo apt-get source --compile aumix sudo dpkg -i aumix_2.8-18_i386.deb sudo dpkg --set-selections
After this last command, I typed "aumix hold", hit return, then hit Ctrl-D. Now the package stays even on an upgrade. Cool.
For some reason, I couldn't see flash websites once I'd installed gutsy. It turns out that I needed to remove the files referring to flash in ~/.mozilla/plugins, remove the package flashplugin-nonfree, and then install it again. Don't know why, but it worked.
Well, I'm trying to get gutsy gibbon onto lighthill, and I'm getting stymied at every turn. First, I tried the commandline version using sudo do-release-upgrade, but I kept getting errors saying "Cannot calculate upgrade" and the log messages showed there was some issue with the gnome-cups-manager package. From google, it seemed that doing the upgrade using the graphical method, gksu "update-manager -c" would take care of this problem, and it looked like it would, but now that method seems to be stuck. I am trying to do this upgrade over ssh, so maybe that's part of the problem. I'm going to go to school now and see what happens from there. I may have to do a new install. Boo.
Okay, at school now, and the gksu "update-manager -c" method seems to be working great. For some reason it was unhappy working over the ssh connection.
Unsurprisingly, the dual-headed-ness didn't come back, but unlike last time, this time I just went to /usr/lib/xorg/modules/drivers and executed
sudo cp mga_drv.so-from_matrox-4.4.3-7.2.0-backup mga_drv.so sudo /etc/init.d/gdm restart
Yay!
I've decided to just downgrade the ATI video driver for now (although I see problems with this in my regular update/upgrading). I downloaded the .deb package with aptitude download xserver-xorg-video-ati and then installed that package with dpkg -i packagename which automatically handled the "downgrade." In order to keep the video driver from being upgraded, I did a dpkg --set-selections and on the next line typed "xserver-xorg-video-ati hold", hit return, then Ctrl-D. Now the upgrade doesn't try to upgrade the ATI video driver. Nice!
The above procedure worked on usagi as well, where I have both an nVidia card and an ATI card. It turns out from the release notes that I'd have to use just the ATI card (if that's even possible) to go to RandR.
EDIT 2007-12-20: To be clear, I had to change the file /etc/apt/sources.list to point to "feisty" instead of "gutsy", then
sudo aptitude update sudo aptitude download xserver-xorg-video-ati/feisty sudo dpkg -i xserver-xorg-video-ati_1%3a6.6.3-2ubuntu6_i386.deb sudo dpkg --set-selections
where I then typed "xserver-xorg-video-ati hold", and hit Ctrl-D. I then editted /etc/apt/sources.list again to change "feisty" to "gutsy," and
sudo aptitude update
Using the reccommended cammandline method, I started the install of gutsy gibbon on my laptop, bleu. Unfortunately, it seems that there were some blocks on the hard drive which had gone bad and caused the process to fail while trying to upgrade ttf-freefont. I used the rescue CD and did fsck.ext3 -fc /dev/sda2 to try to fsck the partition. It found bad blocks and problematic inodes. I let it do all the repairs that it wanted to do.
Now, I was left with the situation that I couldn't seem to upgrade, remove, or reinstall the ttf-freefont package. The error message that I was seeing was "files list file for package `gnome-icon-theme' is missing final newline". Eventually, I needed to repair the file /var/lib/dpkg/info/gnome-icon-theme.list and I did so by following the instructions for FAQ Q5.19 on the Fink documentation page.
Once I'd gotten that done, I could finally reinstall and then upgrade the ttf-freefont package. After that, I still had a weird problem with the new kernel which was installed. When I would boot, X was prevented from starting, and I would see the error message "Re: device-mapper: table: 245:2: linear: dm-linear: device lookup failed." By doing a google search I found info in the ubuntuforums that I simply needed to remove the evms package. After I did that with sudo aptitude remove evms everything seemed to work just fine.
I will note that I haven't tried dual monitors yet, and I have already seen dual monitor problems on the upgrade of my work desktop computer, usagi. There are some notes about this in the release notes for gutsy gibbon.
The recommended commandline method is
sudo do-release-upgrade
It halted with a script error the first time I tried it on usagi, but then I re-ran it, and it worked. Apparently another valid method is
gksu "update-manager -c"
I ran into a problem here and there, primarily with some third party repositories in /etc/apt/sources.list. Also a problem with space on /boot.
I don't know how long it's been out, but I just saw that MoGo is availble for download. I decided to download it and play against it. Well, that turned out to be more of a trial than I expected.
It turns out that glGo (or PANDA-glGo) has some tweaks in it which pretty much tie it to GnuGo as the computer opponent. This kind of sucks.
After searching for quite a while, I finally figured out how to get GoGui to play using MoGo as the computer opponent. Once you've downloaded and unzipped the zip file, go into the gogui directory. In order to get it to install correctly, use the following command.
sudo bash ./install.sh -p /usr/local -j /usr/lib/jvm/java-6-sun
I untarred the MoGo tarball and put the three files (one executable and two data files about openings) into /usr/local/bin. Within GoGui, go to Program -> New Program... Enter "/usr/local/bin/mogo --19 --totalTime 1800" for a game played in 30 minutes (1800 seconds). You may have to mess around with Program -> Attach in order to play MoGo, but that's pretty self-explanatory. The crappy thing is that whichever program one has attached will play both colors, so GoGui cannot be used to watch GnuGo play against MoGo. Ah well.
For some reason, apache2 wasn't configured so I could access the public_html directory in my home directory. It turns out that this is controlled by the Userdir module, which wasn't enabled by default. In order to enable this I used
sudo a2enmod
and from the options given, chose "userdir." Everything works fine now.
EDIT: By also choosing "php5," I avoided the problems with php files in the following configuration.
I have also configured for a localhost only webserver like before.
I received a new (to me) computer from the ME department to use for research. The biggest advantage to this computer is dual processors. Here are the specs.
machine | type | processor | OS | disk | mem | motherboard | location |
---|---|---|---|---|---|---|---|
lighthill | Dell Precision 530 | Dual Intel Xeon 1.80GHz | ubuntu 7.04 | 80G 250G | 1.00G | - | research lab |
I chose to install ubuntu 7.04 from scratch onto this machine. I did run into a weird problem when I removed the second hard drive for a bit and it kept telling me that "primary disc 1" was missing. I was just confused and this refers to the second drive on the primary IDE controller. I was able to turn off the search for it in the BIOS.
I switched the old lighthill to etc4-156-3 for now, and then rsync-ed my home directory over to the new lighthill. Some files couldn't be copied, but they were files which were newer on the new lighthill, so no problem.
In order to get this machine to be the new lighthill, after the install, when I used etc4-156-3 as the temporary hostname, I changed the hostname in /etc/hostname and /etc/hosts. I changed the IP address in /etc/network/interfaces. I also changed some values in /etc/fstab so it would mount the 250G drive in the right place on reboot.
I moved the dualhead Matrox card over to the new machine and followed the basic procedure I had done before to get it running in dualhead mode.
Getting postfix was a little bit of a pain, but mostly because I had forgotten to install procmail and I have that referenced in my .forward. Basically, I installed with "no configuration," and then copied the /etc/postfix directory and /etc/mailname file from the old lighthill.
To get matlab running, I just did a mymakelocalstaff and then copied /usr/local/matlabr14 from the old lighthill to the new.
Packages that I added to the base install so far: openssh-server keychain fvwm m4 mutt fetchmail procmail vim-full gcal apache2 php5 imagemagick mplayer xmms
I got the same computer for a desktop computer at my job. This time, I just swapped the hard drives and DVD burner into the new machine. The dualheaded X configuration didn't work until I went into /etc/X11/xorg.conf and changed the BusID for the second video card to its new address, 4:13.0. Note that lspci listed the address as 4:0d:0, so I needed to translate out of hex.
The network adapter wasn't working either. I was getting the following error.
SIOCSIFADDR: No such device eth0 eth0: ERROR while getting interface flags: No such device
The problem was resolved by commenting out the line in /etc/iftab which connected the MAC address of the previous motherboard's NIC to eth0. Once I commented out that line, the new motherboard's NIC was assigned to eth0. During the research on this problem, the command ifconfig -a helped me to find that the new NIC was being seen at eth1 instead of eth0.
It appears that all of the gnome application and desktop configuration settings are stored in the directory $HOME/.gconf in the directories apps and desktop respectively. I assume copying these between computers will copy settings and configuration. Testing... Neat.
It turns out that copying menus is a different matter altogether. The best way to get something into a menu seems to be through adding a menuitem.desktop file in .local/share/applications. Examples can be found in /usr/share/applications/.
This is handled using the gconftool-2 program. Here are a couple of lines that I used. Note that "stretched" is for fullscreen and "wallpaper" works for tiles.
gconftool-2 -t str -s /desktop/gnome/background/picture_options "stretched" gconftool-2 -t str -s /desktop/gnome/background/picture_filename "filename.jpg"
For quite a while, I've been dealing with some crappy fonts in old X windows programs like cgoban, sanduhr, and even parts of Matlab. I thought that I'd figured out the problem today. It seemed to be with the gtk 1.2 configuration. I fixed this by following the instructions in this ubuntu forum howto. I created the file ~/.gtkrc.mine (referenced from ~/.gtkrc) and in it, I put
include "/usr/share/themes/Plastig/gtk/gtkrc" #include "/usr/share/themes/IndustrialTango/gtk/gtkrc" style "default-text" { # fontset = "-adobe-helvetica-medium-r-*-*-16-*-*-*-p-*-iso10646-*" fontset = "-b&h-lucida-medium-r-*-*-14-*-*-*-*-*-*-*,\ -b&h-lucida-medium-r-*-*-14-*-*-*-*-*-*-*,\ -b&h-lucida-medium-r-*-*-14-*-*-*-*-*-*-*,\ -b&h-lucida-medium-r-*-*-14-*-*-*-*-*-*-*" # fontset = "-bitstream-bitstream vera sans-medium-r-*-*-17-*-*-*-*-*-iso8859-*,\ # -bitstream-bitstream vera sans-medium-r-*-*-17-*-*-*-*-*-iso8859-*,\ # -bitstream-bitstream vera sans-medium-r-*-*-17-*-*-*-*-*-iso8859-*,\ # -bitstream-bitstream vera sans-medium-r-*-*-17-*-*-*-*-*-iso8859-*" } class "GtkWidget" style "default-text"
This fixed some of the problem, but cgoban still didn't look correct. Then I remembered that I can play against GnuGo using glGo, so I decided to do that instead of worrying about how to fix such an old program. I'm sure that there's some "default font" option somewhere in an X-windows config file, but it's not worth tracing down at this point.
I wanted to listen to Harry Potter on my PMP (Personal Music Player), so I loaded it on, and it seemed to be doing fine until I got to the end of the first track, when it seemed to skip a bit. I managed to convinced myself that everything was fine, but then at the end of the next track, it seemed to skip again. This time, I was pretty sure that something was missing. I listened again, and yes it was cut off. I listened to the same track on the computer and it was fine. Bah. (I was becoming more and more dissatisfied with the iriver T10 by the moment, because I found a reference to someone else with this problem, and it only happens in UMS mode. BAH!)
I found the program mp3check, which tests mp3s to make sure that they're within specifications. Well, it turns out that the problem files weren't up to snuff and had about 53k of junk at the start which was causing the rest of the data to get pushed down so that when the player ended at the proper time, the actual stream got cut off. (You can tell I've been listening to a British speaker, no?) Using the command line
mp3check -e --cut-junk-start *
I processed all of the files, and they seem to be playing fine now. Yay! (I can't fault the T10 for only playing mp3 files that meet spec.)
EDIT 2007-08-16:I sent back the Iriver T10. It just ended up not behaving correctly. Too bad, but glad that I got my old PMP working (Cowon Iaudio G3). If I had to buy another PMP right this instant, I'd get a Cowon Iaudio 7.
Well, my Cowon iaudio G3 (1G) has finally started to suffer the effects of all the sweat I pour over it exercising and will randomly turn itself off after 10-20 minutes or so. I've now ordered an Iriver T10 (2G) which I'm convinced I'll be able to turn into a UMS (USB Mass Storage) device so I can use it with linux.
In the meantime, I'm in the middle of Harry Potter and the Deathly Hallows and I don't want to have to wait for the new MP3 player to get here. I have an old Iriver ifp890 (256M) that I could use (though 256M just isn't big enough to use without being a pain). Here's what I had to do in order to get it up and running.
#!/bin/sh # /etc/hotplug/usb/ifpdev chgrp usb $DEVICE chmod g+rw $DEVICE
ifpdev 0x0003 0x4102 0x1108 0x0000 0x0000 0x00 0x00 0x00 0x00 0x00 0x00 0x00000000
/dev/sdb /media/ifp890 auto user,noauto,exec 0 0
Now, when I execute `mount /media/ifp890', the Iriver ifp890 directories are available under /media/ifp890. Nice!
EDIT: Well, I went back, took apart my Cowon iaudio G3, sprayed some DeOxIt into it, and put it back together, and it seems to be working well again. Maybe I didn't need to go and get another MP3 player yet. We'll see.
Well, apparently this upgrade didn't completely finish when I started it. I'm not sure why and I'm not sure why I didn't know. I fixed some problems with the virtual terminals before I even noticed, but it appears that if I'd managed to finish the upgrade, everything would've gone fine.
I did have a problem with grub when I tried to remove some kernel images and set it to "nosplash." Basically, I was using grub-install when I should've been using the update-grub command. I didn't know, but once I figured it out, all the problems with "BIOS disks" not found went away.
This process went well.
There were some minor issues with versions, but I just went with the upgrade options which seemed the most reasonable, then did another "update, upgrade, dist-upgrade" afterwards.
Unsurprisingly, the X server got overwritten, so I followed very similar steps to last time. Getting dual head for my Matrox G450 support back took a bit. First I ended up grabbing the newest version of the driver found at the unofficial matrox linux drivers forum. I then '--extract-only'ed the .run file supplied and grabbed the mga_drv.so and mga_hal_drv.so files from the created xservers/7.1.0 directory and, after turning off X with sudo /etc/init.d/gdm stop, put them in /usr/lib/xorg/modules/drivers/. Because I had already editted /etc/gdm/gdm.conf-custom to add the -ignoreABI flag to the X server, I didn't have to do it again. And finally, I restarted gdm with sudo /etc/init.d/gdm start, and lo and behold, there was dual headed X again. Weee!
I followed the same upgrade procedure above on usagi, and that went well, but the primary screen lost it's resolution information somehow. I went into the /etc/X11/xorg.conf file and uncommented the following lines in the "Monitor" section.
HorizSync 28.0 - 64.0 VertRefresh 43.0 - 60.0
After a restart of gdm, everything looked great.
For some unknown reason, apache2 stopped liking the link in my home directory from public_html to www. It took re-installing apache2 several times before I went to the apache2 error log and finally saw that apache2 really couldn't see it. I removed the link and then recreated the link, and everything worked fine. WTF? I'd wanted to switch to public_html for a while, so I'll do that now. I believe it will affect my home directories, mywwwpush and my web sites at konfuzo.net.
EDIT:It happened again! This time I accidentally changed my "public_html" directory to "public-html" while I was doing some name changes on some .sgf files. Changing it back obviously fixed things. (Thinking back on it, this may have been what I accidentally did last time. Urgh.)
A FAQ in the zsnes forums has the answers to this problem. The solution for me was to kill artsd (which I actually hadn't known I was running, but did allow me to have more than one sound generating program to run at the same time, and switch to 48000 Hz sampling rate.
For the Commodore 64 emulator, x64 (part of VICE), I want X to start in 800x600 mode. As far as I can tell, the best way to do this is to make another "Screen" section in the /etc/X11/xorg.conf file which has only one "Mode" (set to "800x600"). With startx or xinit, one can specify server options after a "--" flag, and one of Xorg's flags is "-screen SCREENNAME". The command line might look something like this.
xinit /usr/local/bin/x64 PAC-MAN.d64 -- -screen x64screen
Will confirm this later today.
Okay, I just wanted to let y'all know that I made some major progress on the MAME machine at this point. In fact, enough progress, that it's no longer the "MAME" machine or even the "arcade" cabinet. This thing isn't even the "game" cabinet anymore.
I managed to get X windows NES and SNES emulators running on the machine. What happens is that when one selects an NES or SNES game on the machine, a script starts X windows, then runs one of these emulators in fullscreen mode in X.
NES and SNES games are very different from the arcade games. The latter only expect 3 minutes of your time for a quarter. The former expect you paid $30 for a game cartridge and then provides dozens of hours of gameplay. This means much more standing in front of the cabinet. I have ordered a gamepad and Bug says he has one to contribute. Aside from allowing more comfortable play on the Nintendo systems, the controllers also open the opportunity to play 4 player games now, like Gauntlet and The Simpsons.
So, on the heels of that success, Kai offered up the challenge of getting the Commodore 64 emulator running on it so he could play C64 games from *his* childhood (the SNES had been Dhruv's childhood). Well, I got that running last night. It's much flakier than the Nintendo emulators, but it does work.
In the process of getting the C64 emulator running, I stumbled across two crucial pieces of information which allowed me to finally get AdvanceMESS (the Multiple Emulator Super System) running. I had to have the BIOS roms for the C64 correctly named and in the right location before AdvanceMESS would stop complaining about "Emulator advmess without roms files, ignoring it." Last night I finally understood that "roms" meant *BIOS* roms and not *cartridge* roms. Now AdvanceMESS will handle C64 emulation, and is now a backup (and crappier) NES and SNES emulator.
Finally, the reason it's no longer necessarily a "game" cabinet is that the C64 is a full fledged computer, on which I entered and ran the following basic program.
10 PRINT "HELLO, KELLYS!" 20 GOTO 10
So, moving forward, it shouldn't be to hard to get other game systems covered by MESS/AdvanceMESS up and running. Anybody need to play those old Apple II games? Atari 2600? Tandy CoCo?
I think we should have an official renaming of the beast, in honor of it's new broadened scope. I thought that the PGEC (pronounced Pee-Geck, the Palace Game and Emulator Cabinet) would be good, but I'm not feeling all that creative at the moment. PGEC rates a 6 out of 10 for naming. I'm sure someone else can come up with a 8 or 9.
Much gaming,
nodog
Here're the important lines from pinkmame:~mame/.advance/advmenu.rc
emulator "zsnes" generic "/usr/bin/xinit" "/usr/local/bin/zsnes %p" emulator_roms "zsnes" "/usr/local/share/advance/image/snes/" emulator "fceu" generic "/usr/bin/xinit" "/usr/games/fceu %p" emulator_roms "fceu" "/usr/local/games/emulation/nes/" emulator "x64" generic "/usr/bin/xinit" "/usr/local/bin/x64 %p" emulator_roms "x64" "/usr/local/games/emulation/c64/"
Looking at these now, I should remark that any command following xinit must begin with a "." or "/".
Here's the important line in pinkmame:~mame/.advance/advmenu.rc.
emulator "advmess" advmess "advmess" ""
Here are the important lines from pinkmame:~mame/.advance/advmess.rc
dir_image /home/mame/.advance/image:/usr/local/share/advance/image:/usr/local/games/emulation dir_rom /home/mame/.advance/:/usr/local/share/advance/bios:/usr/local/games/emulation
In order to get the C64 emulator working, I following both the fine manual and some instructions found in a forum on the web, and acquired the Commodore 64 BIOS roms named c64_basic.zip, c64_kernal.zip, and c64_char.zip which contained the files Basic.rom, Kernal.rom, and Char.rom respectively. AdvanceMESS requires that rom files be named with the number printed on top of the original chip. With some trial and error, and superior feedback provide by AdvanceMESS, I was able to figure out the correct names for these files (Basic.rom -> 901226.01, Kernal.rom -> 901227.03, Char.rom -> 901225.01 ). Those files are `zip'ed together into the file pinkmame:/usr/local/share/advance/bios/c64.zip.
An example commandline to get the c64 emulator running is
advmess c64 -flop PAC-MAN.d64
There is also a zip file I got off the web somewhere (I believe that it was originally called nss.zip), which contains the BIOS roms for the SNES emulator. Those files are in pinkmame:/usr/local/share/advance/bios/snes.zip. The NES emulator apparently doesn't need BIOS roms (?). An example commandline to get the NES emulator running is
advmess nes -cart smbdhtm.nes
I've gotten excited about playing the Super Mario Brothers 1, 2, 3, and Super Mario World on the MAME/MESS cabinet. It's looking more and more like I'm going to switch back to using X on that system, and running xmame and xmess. xmame has been reasonably easy to understand, but xmess took a while to get the syntax right, and the manpage and "-?" option aren't completely helpful. Here's the syntax...
xmess nes -ef 2 -cart SuperMarioBrothers2.nes
which runs the SuperMarioBrothers2.nes rom in the directory /home/nodog/ATTIC/xmess/software/nes. The file /etc/xmame/xmessrc has the following lines in it.
biospath /home/nodog/ATTIC/xmess/bios softwarepath /home/nodog/ATTIC/xmess/software
In order to get it to run a snes program, I got the file nss.zip and renamed it snes.zip and put it in the /home/nodog/ATTIC/xmess/bios directory. xmess isn't a great SNES emulator, though, so zsnes would be better to use.
When I would do a sudo aptitude update on pinkmame, I kept running into a problem with "unknown key"s. It turns out that all I needed to do was sudo aptitude install debian-archive-keyring and then run the update again. I wish I had written down the exact error.
I've been having trouble getting mail through to hotmail, so I thought that I'd try sending mail out through an SMTP server. This is, of course, the correct way to do things, but I haven't had trouble much until now, so I haven't had to worry about it. UT's mail server requires that one connect with SSL for authentication, so I needed to turn that on, which required turning on SASL and TLS. I went through turning on SASL before.
My basic comments for turning on TLS come from the "Getting started, quick and dirty" section of /usr/share/doc/postfix/TLS_README.gz found in the postfix-doc package. I followed the instructions there almost to the letter and it worked, literally the first time.
I'm tired of make-ing my webpages, so I decided to install a local webserver. Installing apache2 was easy (sudo aptitude install apache2), but there were some issues. First, I wanted it to be accessible only from localhost, because I'm not ready to be serving my own web pages yet. I added the following code to /etc/apache2/apache2.conf just before the section about .htaccess.
<Directory /*> Order deny,allow Allow from 127.0.0.1 Deny from all </Directory>
This isn't perfect, because it still allows people to see the names of directories, even if they can't access them, though it's only directories that they knew or guessed the name of. I also wanted php support, so I installed the php5 package (sudo aptitude install php5), and that allowed me to see php files as long as I loaded them from the server (ex. http://localhost/~nodog/academic.konfuzo.net/main/main.php) instead of read them as a file. Finally, I needed to clear my cache on my browser, because otherwise I got the same error message as before installing php5, "you have chosen to open main.php which is a php script."
NOTE: This last error remained persistant on one machine (my home laptop). Eventually I had to install the following symlinks in order to get firefox to load php files (well, to get apache to serve them).
cd /etc/apache2/mods-enabled sudo ln -s /etc/apache2/mods-available/php5.conf sudo ln -s /etc/apache2/mods-available/php5.load
I've been having some troubles getting web movies to play on my browser. A totem screen will appear and say something like "can't play fd0://". Well, it took a bit, but I was able to figure out that I needed to remove the totem-mozilla package and install (or reinstall) the mozilla-mplayer package. Once I did that and restarted firefox, everything worked just fine.
Okay, so I do a bad job of updating my website. I want to create a website that I don't actually need to update each year. This is leading me to thinking about how I use the web, what I publish, what I share, what kind of image I want to have, whether I'm okay with other people holding my data, what I want to be in control of, whether I want to join in the "social" aspect of the web, etc. These are big questions, and not answered easily. This even brings up the dichotomy of what-do-I-want-to-be versus what-I-actually-am.
The first question that I'd like to answer is: How do I currently use my website?
The next question is: How do I currently use the rest of the web in terms of personal website?
I've been beating my head against this huge font problem (menu fonts in firefox, kontact, etc *way* too big) for a couple of hours, and I've finally figured out what the problem was. Before I tell the fix, here are the two lines which eventually gave me the clue I needed to understand the problem.
bleu:~$ xdpyinfo | grep resolution resolution: 105x98 dots per inch bleu:~$ xrdb -q | grep dpi Xft.dpi: 96
The first line tells me what X thinks the dpi of my monitor is. This was originally *way* off. The second shows what Xft thinks my dpi is. If I set this with
echo "Xft.dpi: 105" | xrdb -merge -
Then the fonts were the correct size, but if I added "Xft.dpi: 105" as a line to .Xresources, then it didn't seem to have and effect and in fact
xrdb -q | grep dpi
would return a ridiculous value again (like "175"). I finally read a web page on some fvwm (my window mananger) forum that this had happened to them as well when they started using a dual headed configuration. Well, that set off a lightbulb. I realized that with the "MergedFB" dual headed configuration that I'm using with my ATI card and the open source "radeon" drivers, the X display thinks it's 2800x1050 (both monitors). My first idea was to measure across both monitors and add a line to /etc/X11/xorg.conf that gave the size of both monitors ("DisplaySize 675 270"). A few minutes later, I realized that allowing X to figure out the monitor size on it's own might be a good plan, so I removed the "DisplaySize" lines. Here's the final resulting xorg.conf file with which font display is just fine and Xft.dpi is 96.
# /etc/X11/xorg.conf (xorg X Window System server configuration file) # # This file was generated by dexconf, the Debian X Configuration tool, using # values from the debconf database. # # Edit this file with caution, and see the /etc/X11/xorg.conf manual page. # (Type "man /etc/X11/xorg.conf" at the shell prompt.) # # This file is automatically updated on xserver-xorg package upgrades *only* # if it has not been modified since the last upgrade of the xserver-xorg # package. # # If you have edited this file but would like it to be automatically updated # again, run the following command: # sudo dpkg-reconfigure -phigh xserver-xorg Section "Files" FontPath "/usr/share/X11/fonts/misc" FontPath "/usr/share/X11/fonts/cyrillic" FontPath "/usr/share/X11/fonts/100dpi/:unscaled" FontPath "/usr/share/X11/fonts/75dpi/:unscaled" FontPath "/usr/share/X11/fonts/Type1" FontPath "/usr/share/X11/fonts/100dpi" FontPath "/usr/share/X11/fonts/75dpi" FontPath "/usr/share/fonts/X11/misc" # path to defoma fonts FontPath "/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType" EndSection Section "Module" Load "bitmap" Load "ddc" Load "dri" Load "extmod" Load "freetype" Load "glx" Load "int10" Load "type1" Load "vbe" EndSection Section "InputDevice" Identifier "Generic Keyboard" Driver "kbd" Option "CoreKeyboard" Option "XkbRules" "xorg" Option "XkbModel" "pc104" Option "XkbLayout" "us" Option "XkbOptions" "lv3:ralt_switch" EndSection Section "InputDevice" Identifier "Configured Mouse" Driver "mouse" Option "CorePointer" Option "Device" "/dev/input/mice" Option "Protocol" "ExplorerPS/2" Option "ZAxisMapping" "4 5" Option "Emulate3Buttons" "true" EndSection Section "InputDevice" Identifier "Synaptics Touchpad" Driver "synaptics" Option "SendCoreEvents" "true" Option "Device" "/dev/psaux" Option "Protocol" "auto-dev" Option "HorizScrollDelta" "0" EndSection Section "Device" Identifier "ATI Technologies, Inc. Radeon R250 Lf [Radeon Mobility 9000 M9]" Driver "radeon" BusID "PCI:1:0:0" Option "MergedFB" "true" #Enable MergedFB function Option "MonitorLayout" "LCD, CRT" # Use LCD and CRT even if you have 2 LCD's or CRT's Option "CRT2Hsync" "30-95" #Horizontal Sync of the Monitor (check your monitor's manual for correct values) Option "CRT2VRefresh" "50-160" #Vertical Refresh rate of the Monitor (check your monitor's manual for correct values) Option "OverlayOnCRTC2" "true" Option "CRT2Position" "RightOf" #Physical location of your secondary monitor in relationship to your primary monitor. Option "MetaModes" "1400x1050-1400x1050" #Monitor Resolutions for Primary-Secondary monitors Option "MergedXineramaCRT2IsScreen0" "false" #determines which screen is going to be the primary screen; value can be "true" or "false" EndSection Section "Monitor" Identifier "Generic Monitor" Option "DPMS" # DisplaySize 360 270 # DisplaySize 310 230 # DisplaySize 675 270 # HorizSync 28-70 # VertRefresh 43-60 EndSection Section "Screen" Identifier "Default Screen" Device "ATI Technologies, Inc. Radeon R250 Lf [Radeon Mobility 9000 M9]" Monitor "Generic Monitor" DefaultDepth 24 SubSection "Display" Depth 1 Modes "1400x1050" EndSubSection SubSection "Display" Depth 4 Modes "1400x1050" EndSubSection SubSection "Display" Depth 8 Modes "1400x1050" EndSubSection SubSection "Display" Depth 15 Modes "1400x1050" EndSubSection SubSection "Display" Depth 16 Modes "1400x1050" EndSubSection SubSection "Display" Depth 24 Modes "1400x1050" EndSubSection EndSection Section "ServerLayout" Identifier "Default Layout" Screen "Default Screen" InputDevice "Generic Keyboard" InputDevice "Configured Mouse" InputDevice "Synaptics Touchpad" EndSection Section "DRI" Mode 0666 EndSection
Damn, I tried, but in the end, I can't abide by KDE. The file browser part of konqueror kept crashing on me (it wouldn't do anything, and then suddenly do unexpected things, and I'd have to kill the process). The configuration never ended up working quite right for two screens, and really, the whole thing ended up feeling kind of sluggish, whereas good old fvwm was snappy and behaves as expected.
Unfortunately in that process, I somehow got my fonts in firefox messed up (they were WAY to big), and by that, I mean the fonts in the menus ("File Edit View ..."). The solution was to run gnome-font-properties and in the "Details" button specify 110dpi fonts. That worked like a charm. (Well, worked like a charm for the rest of that session. As soon as I ended my X session and came back in, the fonts got huge again. Urgh.)
Holy bajeebees, that title makes me one of the biggest geeks on earth! But, yes, after I decided to try KDE out again, I realized that it doesn't use xmodmap, but xkb. Well, I didn't want to lose my carefully constructed way of typing Eo characters (ĉĝĥĵŝŭĈĜĤĴŜŬ) using Caps-Lock and Menu as the "Eo shift key" (my IBM laptop has no "win" or "menu" keys, so must have both). Well, I came up with an xkb file to put into /etc/X11/xkb/symbols. It's an exact copy of the file "us" there with the following lines added and I named it "eodv".
// added by nodog 2007-03-08 partial alphanumeric_keys xkb_symbols "dvorak-eo" { name[Group1] = "U.S. English - Dvorak - Esperanto"; key.type = "FOUR_LEVEL"; key <TLDE> { [ grave, asciitilde, dead_grave, dead_tilde ] }; key <AE01> { [ 1, exclam ] }; key <AE02> { [ 2, at ] }; key <AE03> { [ 3, numbersign ] }; key <AE04> { [ 4, dollar ] }; key <AE05> { [ 5, percent ] }; key <AE06> { [ 6, asciicircum, dead_circumflex, dead_circumflex ] }; key <AE07> { [ 7, ampersand ] }; key <AE08> { [ 8, asterisk ] }; key <AE09> { [ 9, parenleft, dead_grave] }; key <AE10> { [ 0, parenright ] }; key <AE11> { [ bracketleft, braceleft ] }; key <AE12> { [ bracketright, braceright, dead_tilde] }; key <AD01> { [ apostrophe, quotedbl, dead_acute, dead_diaeresis ] }; key <AD02> { [ comma, less, dead_cedilla, dead_caron ] }; key <AD03> { [ period, greater, dead_abovedot, periodcentered ] }; key <AD04> { [ p, P ] }; key <AD05> { [ y, Y ] }; key <AD06> { [ f, F ] }; key <AD07> { [ g, G, gcircumflex, Gcircumflex ] }; key <AD08> { [ c, C, ccircumflex, Ccircumflex ] }; key <AD09> { [ r, R ] }; key <AD10> { [ l, L ] }; key <AD11> { [ slash, question ] }; key <AD12> { [ equal, plus ] }; key <AC01> { [ a, A ] }; key <AC02> { [ o, O ] }; key <AC03> { [ e, E ] }; key <AC04> { [ u, U, ubreve, Ubreve ] }; key <AC05> { [ i, I ] }; key <AC06> { [ d, D ] }; key <AC07> { [ h, H, hcircumflex, Hcircumflex ] }; key <AC08> { [ t, T ] }; key <AC09> { [ n, N ] }; key <AC10> { [ s, S, scircumflex, Scircumflex ] }; key <AC11> { [ minus, underscore ] }; key <AB01> { [ semicolon, colon, dead_ogonek, dead_doubleacute ] }; key <AB02> { [ q, Q ] }; key <AB03> { [ j, J, jcircumflex, Jcircumflex ] }; key <AB04> { [ k, K ] }; key <AB05> { [ x, X ] }; key <AB06> { [ b, B ] }; key <AB07> { [ m, M ] }; key <AB08> { [ w, W ] }; key <AB09> { [ v, V ] }; key <AB10> { [ z, Z ] }; key <SPCE> { [ space, space, nobreakspace, nobreakspace ] }; key <CAPS> { type="TWO_LEVEL", [ ISO_Level3_Shift, ISO_Level3_Shift ] }; key <MENU> { type="TWO_LEVEL", [ ISO_Level3_Shift, ISO_Level3_Shift ] }; modifier_map Mod5 { <CAPS> }; modifier_map Mod5 { <MENU> }; };
In order that it would be recognized as a valid keyboard layout, I had to add a line to /etc/X11/xkb/rules/base.lst.
eodv U.S. English w dvorak Eo
in the "! layout" section. Everything worked as expected.
I got a lot of hints about this from this page on custom xkb layouts and Bertilow's page on Esperanto in Linux.
I don't know why but on my research machine, lighthill, some of my matlab windows were using terrible and ugly fonts. These were the dialog pulls up a dialog boxes, and they don't receive the standard font names from Matlab apparently. Well, a solution to this was to add the line
*fontList: -*-helvetica-medium-r-normal-*-17-120-*-*-*-*-iso8859-1
to my .Xresources file (well, in my case to my .Xresources.m4 file). This forces Matlab to choose the specified font. That's pretty okay, but in the process, I think I found out that X in general is choosing a crappy default font. I wonder how I can control that.
Well, now that I have dualheadedness at the research lab and at home, I decided it was time to get it working at my workplace. I didn't expect that they would buy me another video card, so I brought an extra PCI video card from home, an ATI Radeon 7500 (DVI and VGA output). (I had actually thought that this card was broken at one point, but since it's now up and running, I guess that I've proven that it works at least somewhat.) The interesting things about this setup for me is that it's the only place that I've done 2 video cards, and also the only place where my main monitor is on the right. Below is my /etc/X11/xorg.conf.
Section "ServerFlags" Option "xinerama" "true" Option "DefaultServerLayout" "DualHead" EndSection Section "ServerLayout" Identifier "DualHead" Screen "Primary Screen" 0 0 Screen "Secondary Screen" LeftOf "Primary Screen" InputDevice "Generic Keyboard" InputDevice "Configured Mouse" EndSection Section "Files" # path to defoma fonts FontPath "/usr/share/X11/fonts/misc" # FontPath "/usr/share/X11/fonts/cyrillic" FontPath "/usr/share/X11/fonts/100dpi/:unscaled" FontPath "/usr/share/X11/fonts/75dpi/:unscaled" FontPath "/usr/share/X11/fonts/Type1" FontPath "/usr/share/X11/fonts/100dpi" FontPath "/usr/share/X11/fonts/75dpi" FontPath "/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType" EndSection Section "Module" Load "i2c" Load "bitmap" Load "ddc" Load "extmod" Load "freetype" Load "glx" Load "int10" Load "type1" Load "vbe" EndSection Section "InputDevice" Identifier "Generic Keyboard" Driver "kbd" Option "CoreKeyboard" Option "XkbRules" "xorg" Option "XkbModel" "pc104" Option "XkbLayout" "us" EndSection Section "InputDevice" Identifier "Configured Mouse" Driver "mouse" Option "CorePointer" Option "Device" "/dev/input/mice" Option "Protocol" "ExplorerPS/2" Option "ZAxisMapping" "4 5" Option "Emulate3Buttons" "true" EndSection Section "Monitor" Identifier "Dell 1702FP" DisplaySize 330 265 # HorizSync 28.0 - 64.0 # VertRefresh 43.0 - 60.0 Option "DPMS" EndSection Section "Device" Identifier "nvidia Quadro2 Pro 0" Driver "nvidia" BusID "AGP:1:0:0" # Option "IgnoreEDID" "True" # Screen 0 EndSection Section "Device" Identifier "ATI Radeon 7500 RV200 QX" Driver "radeon" BusID "PCI:4:14:0" # Option "IgnoreEDID" "True" # Screen 1 EndSection Section "Screen" Identifier "Primary Screen" Device "nvidia Quadro2 Pro 0" Monitor "DELL 1702FP" DefaultDepth 24 SubSection "Display" Depth 24 Modes "1280x1024" EndSubSection EndSection Section "Screen" Identifier "Secondary Screen" Device "ATI Radeon 7500 RV200 QX" Monitor "DELL 1702FP" DefaultDepth 24 SubSection "Display" Depth 24 Modes "1280x1024" EndSubSection EndSection
xmms started giving me trouble with fonts and doublesizing. The font error was basically "Failed to open font". Here's the doublesize error.
Gdk-ERROR **: BadMatch (invalid parameter attributes) serial 4418 error_code 8 request_code 72 minor_code
The solution to the font errors was simply to go into the preferences and choose a new font, which turned out to be a good thing anyway, because I couldn't really read the previous fonts.
The doublesizing problem apparently has something to do with xmms not choosing a reasonable color depth on doublesizing. The solution is to
export XLIB_SKIP_ARGB_VISUALS=1
before starting xmms. This worked fine, so I added it to my .bashrc.
With this version of matlab on linux, I could compile .c files with the mex command, but when I would run them, I would get errors of the nature
??? Invalid MEX-file '/home/nodog/matlab/AuditoryToolbox/agc.mexglx': /usr/local/matlabr14/bin/glnx86/../../sys/os/glnx86/libgcc_s.so.1: version `GCC_3.3' not found (required by /usr/lib/libstdc++.so.6).
By reading the matlab /usr/local/matlabr14/sys/os/glnx86/README.libstdc++ file and some forums on the net , I saw that I needed to mv the files libgcc_s.so.1 and libstdc++.so.5.0.3 (and the libstdc++.so.5 link to it) somewhere else.
This process went well. I burned an ubuntu-6.10-alternate-i386 cd and followed the instructions here. by mounting the cd at /media/cdrom running gksu "sh /media/cdrom/cdromupgrade". Everything went well, but I lost my dualhead support.
Getting dual head for my Matrox G450 support back took a bit. First I ended up grabbing the newest version of the driver found at the unofficial matrox linux drivers forum. I then '--extract-only'ed the .run file supplied and grabbed the mga_drv.so and mga_hal_drv.so files from the created xservers/7.1.0 directory and, after turning off X with sudo /etc/init.d/gdm stop, put them in /usr/lib/xorg/modules/drivers/. I then had to edit /etc/gdm/gdm.conf-custom to add the -ignoreABI flag to the X server. Here are the exact lines I added.
# Definition of the standard X server. [server-Standard] name=Standard server command=/usr/X11R6/bin/X -ignoreABI -br -audit 0 flexible=true
And finally, I restarted gdm with sudo /etc/init.d/gdm start, and lo and behold, there was dual headed X again. Weee!
EDIT: After the upgrade, I fixed the /etc/apt/sources.list to put my chief repository (ftp.utexas.edu) back in, the did sudo apt-get update, sudo apt-get upgrade, and finally sudo apt-get dist-upgrade. This seemed to go okay, and I did have lots of things which got upgraded.
I followed the same procedure as above, but I couldn't get the "generic" kernel to load. The boot process would stop with "Waiting for root file system. ...". After reading the ubuntuforums, I found that the solution is to rebuild the initramfs for the kernel in question. I did that with the command below.
update-initramfs -k 2.6.17-11-generic -c
NOTE: In both cases above, I also went back into the /boot/grub/menu.lst file and changed the default kernel to either be the generic kernel (with "default 2") or to do the saved/savedefault thing (boot the last kernel booted). To install this, I then executed sudo grub-install hd0.
Well, I went ahead and bought the bullet and upgraded at home as well from CD. This time, however, I chose not to install upgraded packages from the net. That ended up costing me about 20 minutes when I did the update, upgrade, dist-upgrade. Also, the X server went out because that didn't get upgraded along with everything else. It wasn't until I did the update, upgrade that X came back, but X popped up without my intervention. Neat!
This was a nagging problem for a long time: if a local copy of firefox was running, when one 'ssh -X'ed to another machine and ran firefox, then a new window of the local firefox would open up. That was *not* the behavior that I just requested. I wanted to execute the remote copy of firefox. It turns out that there is an environment variable when set to MOZ_NO_REMOTE=1 allows the behavior that I want. (Why would one want the default behavior?!?)
I searched through a bunch of old hardware and determined that it would be possible/good to set up two monitors on my lab machine (lighthill). After beating my head against what I think is a partially broken ATI Radeon 7000 VE, I found a dual monitor Matrox G400 card. It seemed to work great on one head, but gave me a very helpful error of "requires a binary-only "mga_hal" module that is available from Matrox <http://www.matrox.com> ". I went to the page Latest Matrox drivers for older products and grabbed the linux drivers for this card. After untarring the file they gave me, I went into their directory matrox_driver-x86_32-4.4.0/xserver/7.0.0/ and copied mga_drv.so and mga_hal_drv.so to /usr/lib/xorg/modules/drivers/ after first backing up the mga_drv.so which was already in that directory.
Along with the following /etc/X11/xorg.conf file, things seem to be working. There are still a few problems here and there (like where fvwm2 sticks its IconManager and Pager, and there still seems to be a little bouncing on one of the screens), but overall this is a new and neat experience.
# /etc/X11/xorg.conf (xorg X Window System server configuration file) # # This file was generated by dexconf, the Debian X Configuration tool, using # values from the debconf database. # # Edit this file with caution, and see the /etc/X11/xorg.conf manual page. # (Type "man /etc/X11/xorg.conf" at the shell prompt.) # # This file is automatically updated on xserver-xorg package upgrades *only* # if it has not been modified since the last upgrade of the xserver-xorg # package. # # If you have edited this file but would like it to be automatically updated # again, run the following command: # sudo dpkg-reconfigure -phigh xserver-xorg Section "Files" FontPath "/usr/share/X11/fonts/misc" FontPath "/usr/share/X11/fonts/cyrillic" FontPath "/usr/share/X11/fonts/100dpi/:unscaled" FontPath "/usr/share/X11/fonts/75dpi/:unscaled" FontPath "/usr/share/X11/fonts/Type1" FontPath "/usr/share/X11/fonts/100dpi" FontPath "/usr/share/X11/fonts/75dpi" # path to defoma fonts FontPath "/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType" EndSection Section "Module" Load "i2c" Load "bitmap" Load "ddc" # Load "dri" Load "extmod" Load "freetype" Load "glx" Load "int10" Load "type1" Load "vbe" EndSection Section "InputDevice" Identifier "Generic Keyboard" Driver "kbd" Option "CoreKeyboard" Option "XkbRules" "xorg" Option "XkbModel" "pc104" Option "XkbLayout" "us" EndSection Section "InputDevice" Identifier "Configured Mouse" Driver "mouse" Option "CorePointer" Option "Device" "/dev/input/mice" Option "Protocol" "ExplorerPS/2" Option "ZAxisMapping" "4 5" Option "Emulate3Buttons" "true" EndSection # This is for my nvidia card which isn't currently in the machine. #Section "Device" # Identifier "NVIDIA Corporation NV17 [GeForce4 MX 440-SE]" # Driver "nv" # BusID "PCI:2:0:0" #EndSection Section "Device" Identifier "Matrox G400/G450 0" Driver "mga" BusID "PCI:2:0:0" Screen 0 EndSection Section "Device" Identifier "Matrox G400/G450 1" Driver "mga" BusID "PCI:2:0:0" Screen 1 EndSection Section "Monitor" Identifier "Viewsonic A90 Monitor" VendorName "Viewsonic" ModelName "A90" HorizSync 30-86 # These are the actual Vert rates for the Viewsonic A90 Monitor. # VertRefresh 50-180 # The rates below are faked to have two synchronized monitors. VertRefresh 50-160 Option "DPMS" EndSection Section "Monitor" Identifier "Dell M992 Monitor" VendorName "Dell" ModelName "M992" # These are the actual Horizontal rates for the Dell M992 Monitor. # HorizSync 30-96 # The rates below are faked to have two synchronized monitors. HorizSync 30-86 VertRefresh 50-160 Option "DPMS" EndSection Section "Screen" Identifier "Default Left Screen" Device "Matrox G400/G450 0" Monitor "Dell M992 Monitor" DefaultDepth 24 SubSection "Display" Depth 1 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 4 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 8 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 15 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 16 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 24 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection EndSection Section "Screen" Identifier "Right Screen" Device "Matrox G400/G450 1" Monitor "Viewsonic A90 Monitor" DefaultDepth 24 SubSection "Display" Depth 1 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 4 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 8 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 15 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 16 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection SubSection "Display" Depth 24 Modes "1280x1024" "1152x864" "1024x768" "832x624" "800x600" "720x400" "640x480" EndSubSection EndSection Section "ServerLayout" Identifier "Dual Head Layout" Screen "Default Left Screen" Screen "Right Screen" RightOf "Default Left Screen" InputDevice "Generic Keyboard" InputDevice "Configured Mouse" EndSection Section "ServerFlags" Option "xinerama" "true" Option "DefaultServerLayout" "Dual Head Layout" EndSection Section "DRI" Mode 0666 EndSection
Following the success of what I'd done at the lab and seeing that it might be possible to use the ATI Radeon R250 Lf fire GL 9000 (Mobility 9000 M9) video card on my IBM Thinkpad R50 to run the regular laptop screen at the same time as an external monitor, I tried it. After a couple of hours of fiddling with it, it worked. It's using xinerama, which apparently isn't optimal, but works for now. I did have to download a proprietary driver for this video card, the fglrx driver, but instructions on how to do that were given here. Here's my /etc/X11/xorg.conf.
Section "ServerFlags" Option "xinerama" "true" Option "DefaultServerLayout" "DualHead" EndSection Section "ServerLayout" Identifier "DualHead" Screen "laptopScreen" Screen "externalScreen" RightOf "laptopScreen" InputDevice "Generic Keyboard" InputDevice "Configured Mouse" InputDevice "Synaptics Touchpad" EndSection Section "Files" # path to defoma fonts FontPath "/usr/share/X11/fonts/misc" # FontPath "/usr/share/X11/fonts/cyrillic" FontPath "/usr/share/X11/fonts/100dpi/:unscaled" FontPath "/usr/share/X11/fonts/75dpi/:unscaled" FontPath "/usr/share/X11/fonts/Type1" FontPath "/usr/share/X11/fonts/100dpi" FontPath "/usr/share/X11/fonts/75dpi" FontPath "/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType" EndSection Section "Module" Load "i2c" Load "bitmap" Load "ddc" # Load "dri" Load "extmod" Load "freetype" Load "glx" Load "int10" Load "type1" Load "vbe" EndSection Section "InputDevice" Identifier "Generic Keyboard" Driver "kbd" Option "CoreKeyboard" Option "XkbRules" "xorg" Option "XkbModel" "pc104" Option "XkbLayout" "us" EndSection Section "InputDevice" Identifier "Configured Mouse" Driver "mouse" Option "CorePointer" Option "Device" "/dev/input/mice" Option "Protocol" "ExplorerPS/2" Option "ZAxisMapping" "4 5" Option "Emulate3Buttons" "true" EndSection Section "InputDevice" Identifier "Synaptics Touchpad" Driver "synaptics" Option "SendCoreEvents" "true" Option "Device" "/dev/psaux" Option "Protocol" "auto-dev" Option "HorizScrollDelta" "0" EndSection Section "Monitor" Identifier "laptopMonitor" DisplaySize 305 228 Option "VendorName" "ATI Proprietary Driver" Option "ModelName" "Generic Autodetecting Monitor" Option "DPMS" "true" EndSection Section "Monitor" Identifier "externalMonitor" DisplaySize 370 275 Option "VendorName" "ATI Proprietary Driver" Option "ModelName" "Generic Autodetecting Monitor" Option "DPMS" "true" EndSection Section "Device" Identifier "atiRadeonR250Lf_Mobility9000M9-0" Driver "fglrx" Option "VideoOverlay" "on" Option "OpenGLOverlay" "off" BusID "PCI:1:0:0" Screen 0 EndSection Section "Device" Identifier "atiRadeonR250Lf_Mobility9000M9-external" Driver "fglrx" Option "VideoOverlay" "on" Option "OpenGLOverlay" "off" BusID "PCI:1:0:0" Screen 1 EndSection Section "Screen" Identifier "laptopScreen" Device "atiRadeonR250Lf_Mobility9000M9-0" Monitor "laptopMonitor" DefaultDepth 24 SubSection "Display" Viewport 0 0 Depth 24 EndSubSection EndSection Section "Screen" Identifier "externalScreen" Device "atiRadeonR250Lf_Mobility9000M9-external" Monitor "externalMonitor" DefaultDepth 24 SubSection "Display" Viewport 0 0 Depth 24 EndSubSection EndSection Section "DRI" Mode 0666 EndSection
I had to create a new Mac today with OSX so that a labmate could replace the Windoze machine that a professor needed back. I just yanked the OSX drive out of the mac that I'd created, put the drive in another G4 500MHz that was sitting around, booted with the option key, selected the OSX partition (the only one), went into System Preferences->Startup Disk, chose that partition, and rebooted. That machine is fine.
The old ubuntu drive, though, needed to be switch from slave to master, and that proved to be trouble. The end result/solution was to use the ubuntu-6.10-alternate-powerpc disc and choose "rescue" at the prompt. Eventually, I was able to start a shell with partition hda3 as root, edit /etc/yaboot.conf to reference "hda" instead of "hdb" and "disk0" instead of "disk1" and then run ybin. The disk0/disk1 bit was tricky to find/see at first, and caused some extra work. On the plus side, I did get to know more about OpenFirmware and how that works.
EDIT 2007-02-16: At some point, I added a second disk as a slave back to the G4 running ubuntu so other students in the acoustics lab would be able to use a machine with OS X on it. After the OS X install, of course the machine booted directly into OS X. Getting it to boot at least into yaboot again was pretty simple. I just held down the "option" key on boot and selected the linux drive for boot. After booting, I used fdisk on the second drive to find out which partition OS X was on (/dev/hdb5), added that to the /etc/yaboot.conf file (macosx=/dev/hdb5) and a "defaultos=macosx", and then ran ybin -v again. Yay!
I was trying to run qiv on ubuntu 6.10 Edgy Eft on an old G4 PowerPC, but I kept getting weird errors of "Gdk-ERROR **: BadMatch (invalid parameter attributes)". It turns out that this is related to "Compositing" in X. There are two workarounds which seem to handle the problem.
Section "Extensions" Option "Composite" "Enable" EndSection
The second is probably preferable, but I used the first for now.
It took me a while but I finally figured out how to stop xscreensaver on ubuntu 6.06 from locking the screen when I closed the lid. The file /etc/default/acpi-support contains these two lines.
# Comment this out to disable screen locking on resume LOCK_SCREEN=true
So, I just commented the second line. It may not take effect until the next time I login or restart the system, though.
I figured out that the reason that the mouse wasn't working on our MAME cabinet was that user mame didn't have any permission on
/dev/input/event[0-9]* /dev/input/mice /dev/input/mouse[0-9]*
In order to give read and write permission, I first added mame to group games. I then added the following lines to /etc/udev/rules.d/40.permission-rules.
KERNEL=="event[0-9]*", GROUP="games" KERNEL=="mice", GROUP="games" KERNEL=="mouse[0-9]*", GROUP="games"
and on a /etc/init.d/udev restart all of the relevant files in /dev/input are now group games.
Yay! Trakball on MAME so I can play Tempest!
EDIT: Okay, so the above was done on an ubuntu 6.06 system, and the system I really wanted to use was a debian etch system. This meant that the file I added the lines to was /etc/udev/rules.d/020_permission.rules and a reboot was required (perhaps not required, but it worked).
I had an extra 80G disk laying around that I know had some performance problems on a high end server that doesn't allow performance problems. I thought that would make a great home directory disk on sedusa. I installed it as the slave on the second IDE controller. I then booted into "recovery mode" which appears to simply be single-user mode. I used fdisk to give it the following partitions:
hdd1 swap 2G 82 hdd2 / 12G 83 hdd3 /boot 500M 83 hdd4 /home 65G 83
In order to set up the filesystems, I ran the following:
mkfs -V -t ext3 -c -m1 /dev/hdd4 mkfs -V -t ext3 -c -m1 /dev/hdd3 mkfs -V -t ext3 -c -m1 /dev/hdd2
There were some errors on the second command, but I hope this was just the mkfs program recognizing the bad sectors and marking them.
I temporarily mounted partition hdd4 to /mnt with mount /dev/hdd4 /mnt and used cp -dpr /home/* /mnt to copy the home directories over.
Finally, I changed the line in /etc/fstab to mount /dev/hdd4 as home instead of /dev/hda4:
/dev/hdd4 /home ext3 defaults 0 2
and rebooted. Everything came up just fine and looks good.
ssh remotehost /usr/games/gnugo --level 6 --mode gmp --quiet --cache-size 32
cgoban would keep beeping and tell me "waiting for go modem handshaking to complete." I finally figured out that ssh wasn't in cgoban's PATH, and gave it the full pathname.
/usr/bin/ssh remotehost /usr/games/gnugo --level 6 --mode gmp --quiet --cache-size 32
EDIT: Well, there still seem to be other problems. Feh.
reEDIT: Well, this actually seems to work now.
I've been miffed because when I look at programs on different computers, the physical size of the fonts on my screen will look very different, even though I specified the same font size in the program and, to my knowledge, I'd chosen the same parameters for the different machines when setting them up. This even occurred for displaying a program remotely. It might look great on one computer, but remotely displaying it on another made it look terrible due to font size problems.
Thankfully, I found the solution today at http://www.mozilla.org/unix/dpi.html. Basically, one needs to use the "DisplaySize" command in the "Monitor" section of /etc/X11/xorg.conf to specify the actual size in mm of the viewable area of one's monitor. With a full restart of X, this made everything much better. Yay!
The specific line in my /etc/X11/xorg.conf looks like.
DisplaySize 330 265
From ubuntuguide.org
sudo apt-get install libdvdread3 sudo apt-get install dpkg-dev fakeroot debhelper # If you have AMD64. You need dpkg-source. Otherwise skip. sudo /usr/share/doc/libdvdread3/examples/install-css.sh
Everything seemed to be working fine with gouda, but on an apt-get update && apt-get upgrade followed by a reboot my orinoco silver wireless card wasn't working anymore. To get networking up again at all, I installed the old 3Com wired PCMCIA card that I had (Dell FastEthernet 10/100), started X, and used gnome's network-admin program (which seems pretty good). After searching around on the web, I found this page which had a suggestion which worked for me. It turns out that the "hostap" and "orinoco" drivers were battling it out for the orinoco card. I added
blacklist hostap_cs
to /etc/modprobe.d/blacklist, rebooted, and everything was great. Yay!
I use gouda (the Dell Inspiron 7000) as my stereo computer (it plays my music). In order to keep it connected to the network, the screen has to stay at least slightly open. If the backlight is one, that puts too much light into the room, so I needed to shut off the backlight. It turns out that was a two step process.
First I needed to recompile the kernel with the APM_DISPLAY_BLANK option turned on. I also followed some suggestions online for speeding up the kernel. That's in the next section.
Second I needed to somehow effect the same output as setterm -blank 5 on the virtual console without anyone logging in. It turns out that command actually outputs an escape sequence, so setterm -blank 5 > blank_5.uxt stored that escape sequence in the file blank_5.uxt. getty is the program that puts the login prompt on the screen as well after it shows the contents of /etc/issue. So, I just dumped the contents of blank_5.uxt into /etc/issue on a separate line, and now the screen blanks after 5 minutes.
Here are a few tips I found on the web for speeding up a 2.6.15 kernel...
In "General Setup" activate: -Support for paging of anonymous memory (swap) --Support for prefetching swapped memory In "Processor type and features": -Processor family Choose the model of your processor. Activate: -Preemption Model --Voluntary Kernel Preemption (Desktop) -High Memory Support --off -if you have less than 1 GB of RAM --1GB Low Memory Support -if you have 1GB of RAM --4GB -if you have more than 1GB of RAM -Timer frequency --1000 Hz In "Device drivers" go to "Block devices" and in "IO Schedulers" leave only the "CFQ I/O scheduler" activated, which provides the best performance. In "Kernel hacking" uncheck "Kernel debugging".
In a fit of useless activity, I decided that my Dell Inspiron 7000 (gouda) needed to no longer be running SuSE 9.3 and needed to be switched to debian or ubuntu. I figured that debian's known for being a distribution which works well on old hardware (here a PII-333). I only use the machine to play music to my stereo anyway (so why did I put any effort into this at all?).
So, I kept getting an error while trying to install either debian or ubuntu. (I'll add the exact text of the error later.) It turns out that on starting PCMCIA card services, there was one set of ports and irqs that I needed to exclude for the CDROM to work completely correctly, and another set of ports that I needed to include for the orinoco silver wireless network card to work at all. Here's the final version of the line I needed to type into the "config" line while starting card services.
exclude port 0x800-0x8ff, irq 3, irq 5, include port 0x300-0x4ff, port 0x000-0xcff
In the end, I decided to install ubuntu. The biggest reason, was that all of my other machines are ubuntu, so the /etc/apt/sources.list file will be the same, and I can expect similar behavior. The biggest drawback will be the installation of far more packages than I need on this simple machine. I'm gonna have to get rid of some packages and turn X off, but that'll be fine as long as there is enough room to install the full distribution.
This did force me to get wireless working on this machine, which I did want.
I don't love the ubuntu boot splash screen with it's little progress bar, so I wanted to disable it. It was just a matter of removing "quiet" and "splash" from the boot line in /boot/grub/menu.lst, and then running sudo grub-install /dev/hda. It worked out great, although I could switch to a more modern vga mode, but that's much less necessary.
Debian, and therefore ubuntu, makes use of /etc/alternatives to choose which of a group of programs which fills a particular role will be chosen when the generic name for that role is invoked. As an example, which of x11-ssh-askpass, gtk-led-askpass, ssh-askpass-gnome, ssh-askpass-fullscreen will be chosen when the user types in ssh-askpass. There are lots of command line options to make things happen immediately, but the easy route is just to type in
sudo update-alternatives --config ssh-askpass
and select one of the choices by number.
I finally figured out the correct URI to give to CUPS in order to print from one CUPS server to another ("printers" is literal here, "machine.domain" and "queue" are the names of the machine and the printer queue):
ipp://machine.domain/printers/queue
By reading up on this, I did avoid another problem, which was that when printing from one cups server to another, one must set one of the servers to "raw" as the printer manufacturer and type. This avoids "double translatting" the file.
I will note that it is very helpful if *you* are a member of the "lpadmin" group and user "cupsys" is a member of the "shadow" group. (You'll need to /etc/init.d/cupsys force-reload to get the running cupsys to actually be in the shadow group.) If so, then one can use one's own username and password instead of root when adding a user. This is especially useful when on ubuntu, because there is no root user.
From shallowsky.com I found that the "Enter username and password for 'CUPS'" dialog can actually be handled by doing the following:
adduser cupsys shadow adduser yourname lpadmin
The second line is only necessary if you are not a member of lpadmin already. This was indicated in the very end of /usr/share/doc/cupsys/README.Debian.gz
If you copy a URL wrapped over multiple lines from somewhere and try to paste it into the address bar, you will end up only with the first line of it. To fix it, go to about:config and change editor.singleLine.pasteNewlines setting to 3.
In order to paste a URL (middle-mouse button) into a Firefox browser window or tab does and have that cause the page to be opened: Open about:config (by writing it in the location bar) and set middlemouse.contentLoadURL to true.
So, xbiff's pixmaps have reverted to the terrible looking inbox images and I was getting these errors in .xsession-errors:
Warning: Cannot convert string "flagup" to type Pixmap Warning: Cannot convert string "flagdown" to type Pixmap
It turns out that the bitmaps for xbiff are contained in the "xbitmaps" package, and that needed to be installed.
I took a look at many different sources.list files for ubuntu, and this is the conglamoration that I came up with:
## Add comments (##) in front of any line to remove it from being checked. ## Use the following sources.list at your own risk. ## deb http://archive.ubuntu.com/ubuntu dapper main restricted universe multiverse ## deb-src http://archive.ubuntu.com/ubuntu dapper main restricted universe multiverse ## Main UT Mirror deb ftp://ftp.utexas.edu/mirrors/ubuntu/ dapper main restricted universe multiverse deb-src ftp://ftp.utexas.edu/mirrors/ubuntu/ dapper main restricted universe multiverse ## MAJOR BUG FIX UPDATES produced after the final release deb http://archive.ubuntu.com/ubuntu dapper-updates main restricted universe multiverse deb-src http://archive.ubuntu.com/ubuntu dapper-updates main restricted universe multiverse ## UBUNTU SECURITY UPDATES deb http://security.ubuntu.com/ubuntu dapper-security main restricted universe multiverse deb-src http://security.ubuntu.com/ubuntu dapper-security main restricted universe multiverse ## BACKPORTS REPOSITORY (Unsupported. May contain illegal packages. Use at own risk.) deb http://archive.ubuntu.com/ubuntu dapper-backports main restricted universe multiverse deb-src http://archive.ubuntu.com/ubuntu dapper-backports main restricted universe multiverse ## PLF REPOSITORY (Unsupported. May contain illegal packages. Use at own risk.) ## requires the following line to add the repository key... ## wget http://packages.freecontrib.org/ubuntu/plf/12B83718.gpg -O- | sudo apt-key add - deb http://packages.freecontrib.org/plf dapper free non-free deb-src http://packages.freecontrib.org/plf dapper free non-free ## CANONICAL COMMERCIAL REPOSITORY (Hosted on Canonical servers, not Ubuntu ## servers. RealPlayer10, Opera and more to come.) deb http://archive.canonical.com/ubuntu dapper-commercial main ## Repository for Skype ## deb http://download.skype.com/linux/repos/debian/ stable non-free
Okay, so there are several packages that are in debian stable and in debian unstable that are missing in debian testing (notable vlc and mplayer). What a pain. The final solution to this has now been to switch back to ubuntu. Jeebus! I'm getting all of my machines there slowly, but the pink palace machine sedusa freaked out during the ubuntu install and won't even get to the grub menu. "Error 17"
Following this page I was able to get postfix to do sasl authentication with mail.konfuzo.net fairly easily.
relayhost = [mail.konfuzo.net] smtp_sasl_auth_enable = yes smtp_sasl_password_maps = hash:/etc/postfix/smtp_pass smtp_sasl_security_options =
Note that the "[]" around mail.konfuzo.net tells postfix to not use MX rocords, but A records instead. This is apparently "the trick."
mail.konfuzo.net: myusername:mypassword
And it seems to be working!
I don't completely understand the debian/cdrecord/wodim problems, but I couldn't get cdrecord to work properly with debian. Eventually I found that I needed to run modprobe ide-cd and set CDR_DEVICE=/dev/cdrw
So, it was quite strange, but my during an Online Update on my SuSE machines, I've been getting "No patches available" when I know that patches have been released. It turns out that I need to go to "Online Update Configuration" and "configure now." Then everything seems to be in order again.
It took a while before I found the website http://easymamecab.mameworld.net/, but once I did, I finally figured out that the trick to getting a framebuffer kernel that wasn't VESA was to remove the VESA driver and compile the framebuffer driver for your video card into the kernel instead of as a module. Now I've got the MAME machine using AdvanceMAME through debian. Now the big problem is the A-PAC controllers. They really seem to see left and right on the joysticks as the same thing. (Up/down as the same, too.)
I was trying to figure out which video card I had in the machine when my boss suggested the lspci command. It gave me everything I needed.
I've been beating my head against the foreign characters being printed when using Alt-anything in xterm for a little while. I finally found the solution. I had to turn off 8-bit input. This was done by adding a line with "xterm*eightBitInput: false" in it to .Xresources (well, in my case, to .Xresources.m4 because I process that file with m4 in mycrazify). In order to do debugging, I used xrdb -merge ~/.Xresources to immediately put the changes into effect. Now, I can use irssi on my debian and ubuntu machines.
So, for better or for worse, I've decided to put ubuntu 6.06.1 LTS on my laptop instead of SuSE 10.1. It seems like I'll be doing a little more sysadm work to get things set up, but hopefully, apt-get and a speed increase will be the result. So far, I'm a little dubious, but still moving forward. (Note: In the future, I would recommend using the "alternate" install for anyone except rank beginners. The "Live CD" install is pretty, but doesn't allow for many options.)
The biggest problems I've had so far are
1 - I didn't realize that I hadn't installed procmail. This caused my .forward file (consisting of one line: "| /usr/bin/procmail || exit 75") to cause mail delivery to fail. I thought there was a problem with my exim4 configuration, so I reinstalled it several times. Eventually, I gave up and installed postfix instead, and thankfully, it showed me my error almost immediately.
2 - I'm getting foreign characters when I hold down alt and type in an xterm. This wouldn't be a big deal, I guess, except that I used alt-1, alt-2, etc to get to the different IRC windows in irssi. This new behavior prevents access to those windows, except through the commands /window 1, /window 2, etc. rxvt and other programs don't exhibit this behavior.
The wireless network card gave me a little bit of trouble, but I think that once I typed all the config info in correctly, the rest went smoothly. I'll also note that network-admin sure makes wireless connections easier!
So, I need to authenticate at konfuzo.net before it will relay my mail. I've set this up before, following instruction found at wiki.dreamhost.com. It looks like it's easier to set up an authenticated smarthost using exim. I followed the instructions at http://wiki.dreamhost.com/index.php/Exim, but then later was able to remove my /etc/exim/passwd.client file and konfuzo.net seems to keep relaying. I'm not sure why. Maybe it's a IMAP before SMTP thing. I dunno.
After some decision making at work, we've decided to go with debian instead of ubuntu, so I decided to do the same thing at home. Once I'd moved my /home directory to a separate partition (which I will advocate from here on) I was able to reinstall an operating system in about 25 minutes. Getting wireless working (simply a matter of putting the ipw2100-1.3.fw (found by google) in /usr/lib/hotplug/firmware) and getting mail working (which involved me putting my mail.konfuzo.net username and password in /etc/exim4/passwd.client) were the only two real hold ups. I have to admit that I love apt-get and how easy it is to immediately install a package that you didn't have. Much nicer than waiting minutes for it with SuSE and YaST. (Oh, and network-admin in the gnome-system-tools is just badass for working with wireless networks. I was able to switch back and forth between school and home networks super simply.)
Apparently, for a non-initrd kernel, you need to have the following compiled into your kernel: motherboard chipset driver, driver for the boot drive type and controller, and a driver for the root filesystem.
Okay, so I suddenly find myself looking at debian again, both for work and for a MAME machine. Here are my notes about compiling an etch (testing) kernel (2.6.16) for debian.
VFS: Cannot open root device "hda1" Please append a correct "root=" option Kernel panic: VFS: Unable to mount root fs on unknown-block(0,0)
It's up and running after several tries. Now, I'd like to create a kernel that doesn't depend on an initrd image as well as has the options that I want. Moving along.
SuSE seems to have gotten rid of it's automounting (at least as default) of USB drives, so one must mount them. Here's what I did to get my USB drive into my own home directory.
sudo mount -o uid=513 /dev/sda1 ~/mnt
mount -t smbfs -o username=nodog,password=whatever,ip=146.6.83.XXX //blah/dump1 ~nodog/mnt
The "blah" is literal (I don't think it matters because the IP is specified), but the "XXX" needs to be replaced.
I'm just looking to make an ISO image directly from the CD instead of mounting it and mkisofsing it. The solution is dd. First, stick in the cd, then umount it, then
dd if=/dev/cdrom of=outfile.iso
Pretty simple, no?
I've been working with a Palm V, trying to use it as a go recorder, and it's been a bit of a pain in some ways, but easy in others. Just getting the Palm connected to the linux box has been pretty easy.
sudo chmod 0666 /dev/ttyUSB0 sudo ln -s /dev/ttyUSB0 /dev/pilot
and use the program pilot-xfer. The next step is to actually get information out of the MemoDB on the pilot.
pilot-xfer -f MemoDB
Then it's actually easiest to just edit this file and copy-n-paste to another file. Another option is to use the memos program to make the output look a little nicer.
memos -f MemoDB.pdb > outputfile
I think that's about all you need.
-----BEGIN PGP PUBLIC KEY BLOCK----- Version: GnuPG v1.4.0 (GNU/Linux) mQGiBEQ1icoRBAC2Yz733P8f+IpS3jqXktw9Fs87M1WtHqLjt0z74KdKp6X2U8Oz APr6QXHlB9GgNugGu4vnxQqW4i9YpuV2r9l0AgxgNBx+KCdNYAr4jFdvrIFoh4hO bbC/WKtkwjf7C4UPhEkGj23F4WkRKSvPYv5jQHlGk40ofac4ersY08uC0wCg8/yc hotGvYFIvfYZcRm2IQbmn/UEAI8fh9hqIYtUQJeAQWzsnQ11qWsa6VdCZlebnpNi j2prOi3KFw/68MjDangbEbKjsV1RVYN4CLTfXxilKpeegRfp+L+3mm//G4ggjFEe dz/xTPQEM7XKY5JYFxeaYa0WEKj4oEnA6OF12n0q4KvU2lfbuyXStAggYI6RBbNc 2sFMBACEGw/DkWh9SzPyrBz8wxwJQNF9kUj5XFOHIYBMqHfg5Nfah/Q9Okyyn+gD 732zydLC9QrDhYXTUV/i8qFctVFY7jJw3bPHctnGqCEICBSOYfjVdH9z1c1gy+Xz Jw2OiFE0RefujAXYkzQ+VQq5KulFNU9qca8iXRNSzyxnBR3QprQtQW5kZXJzb24g TWlsbHMgKEpvaG4gSUlJKSA8bm9kb2dAa29uZnV6by5uZXQ+iF4EExECAB4FAkQ1 icoCGwMGCwkIBwMCAxUCAwMWAgECHgECF4AACgkQE+tFcI7NNYPWkwCdHmoI5+MR KKORZbOEdxXKNlmIFdkAn2bhpLAE5TKocQVxzr3uT+1Uuk7guQINBEQ1idQQCADk 7OtKb8zmqHserqyByj91jH7ncq1CDtWCEbVO4RE1RZ1/Qgbo4Jzg3BMe9kP6sMe5 sPzcKzAxgHWrC2o6xgMHIvJkbfjJPqk66znPILVtY+o1MLVFa5NVoRy+xvgLGuKm d08xM2oG6XIH7TPVKQ5UPsBNeXivEI25GUpm6Sv9BzrPYgaPHETiLkda40ZlUiZ4 /VT+qXSHFEbq3fgj472p4koS7RfsBCAk7vB5HxFTus/7YWlRdbNUPk5Zqw1UqnOB JvNu6PDqR8rQQkMnaXqKhd5i8cHDC9WMBIkiDc5iTMTGz9zBPHe3IaPzWR9QnsCC e+JMEc3zEvsQvKR8uJ7LAAMFB/0eZBo+qlX4b70fxo7vGeacan2qQU1qZ7+M5mXq 8qN2u1d0Sf03mv0XMRcBQPO8i9ult128Gq6dUZ4iJiZ+xjoMhj9vLp8fFoG0ILmV YeXnI6wyeeNiqzyJzDav/PdXuBswM1x08EF535Dcjiqt32BN6EVZp99Ju2SvKbfm 5CYcDL2224Z7zJY4P58WRrnrH6WJs/omBGhsoWvZczUuiAARaXR+Ni6YHijCMa2p 16uEECHq8BxOEsWcngQCr74VQCrqs2kF7nmyRSshvwo1ylOUkdMMAz4Q9MFtzt8j 1CMb8LhOsLDLX+xZHzHKdHNp2h00A1Gnk9EOckmUK+trv5criEkEGBECAAkFAkQ1 idQCGwwACgkQE+tFcI7NNYNZGwCg3t4MpuCq1ntZ74ZflqYYjNa1wWUAoJXXyc86 eezilVe9LVxd3n+7b550 =rYgX -----END PGP PUBLIC KEY BLOCK-----
pub 1024D/8ECD3583 2006-04-06 Key fingerprint = F91D 1BEA CEB1 93D1 5B08 A7DF 13EB 4570 8ECD 3583 uid Anderson Mills (John III)sub 2048g/02ADA6C9 2006-04-06
Just trying to re-encode a BIG file to be smaller...
nice -19 mencoder -vf scale -zoom -xy 480 -oac copy -ovc lavc -lavcopts vcodec=mpeg4:vbitrate=700:vhq:vpass=1:turbo -o outfile.avi infile nice -19 mencoder -vf scale -zoom -xy 480 -oac lavc -ovc lavc -lavcopts vcodec=mpeg4:vbitrate=700:vhq:vpass=2:acodec=mp3:abitrate=128 -o outfile.avi infile
So, I suddenly had trouble with ripping CDs on my lab machine, lighthill. I was trying to get grip to rip, but it would immediately think that the ripping was done and then try to encode non-existent files. I changed two things at once and it started working. I had already changed the cdrom device to /dev/sr0 (my first SCSI device (since my cdrom is a scsi device)), but then I added g+w permissions to /dev/sr* and /dev/sg* (sudo chmod g+w /dev/sg* /dev/sr*) while I named /dev/sg0 as the generic SCSI device to cdparanoia in grip. Life was grand once again.
So, my portable music player suddenly couldn't talk to my linux system when I loaded SuSE 9.3. I don't know what exactly the problem was, but the fix was to make this change in /etc/fstab:
usbdevfs /proc/bus/usb usbdevfs noauto 0 0needs to be changed to
usbfs /proc/bus/usb usbfs noauto 0 0and reboot.
Yay, I can see my iriver again, and use ifp-line.
Today I wanted to re-encode a big file to make it smaller...
mencoder -vf scale -zoom -xy 512 -endpos 0:21 -oac lavc -ovc lavc -lavcopts vcodec=msmpeg4:vbitrate=800:vhq:vpass=1 -o kmpgp.avi Why_I_Wear_Big_Pants.mp4 mencoder -vf scale -zoom -xy 512 -endpos 0:21 -oac lavc -ovc lavc -lavcopts vcodec=msmpeg4:vbitrate=800:vhq:vpass=2 -o kmpgp.avi Why_I_Wear_Big_Pants.mp4
I was trying to compile an older LaTeX document which had some unicode characters in it, and I got an error during the compile:
Package inputenc Error: Unicode char \u8:ĉ not set up for use with LaTeX.It turns out that an extra copy of utf8.def is now located in /usr/share/texmf/tex/latex/base that was causing the problem. I just changed it's name to utf8-moved.def, and everything works fine again. Apparently this is a problem with SuSE 9.3 and was discovered by these Germans.
I've upgraded both bleu and lighthill to SuSE 9.3. Actually, I did a complete install on lighthill, so it's very fresh smelling now.
So, I couldn't get mplayer to play from a USB DVD recorder, because I couldn't figure out the correct device name. Finally, I realized if I let SuSE automount a data CD, and then looked at the device name, I'd have it. It turned out that it was /dev/sr0, and the correct command line was
mplayer dvd://1 -dvd-device /dev/sr0
Aha! Also, it only works if I do this command line immediately after inserting the DVD into the drive.
So, I'm trying to create Esperanto subtitles for a movie which is on a DVD. I'm using mencoder to extract the VOBsub files. I did this with the command line
mencoder dvd://1 -dvd-device /dev/sr0 -vobsubout mysubtitles \ -vobsuboutindex 0 -vobsuboutid en -sid 1 -nosound -ovc copy
After that, I used the set of programs called subtitleripper. Apparently these need to be put into the contrib/suprip directory of transcode (though I couldn't tell if this was absolutely necessary). At this point, I've gotten the program vobsub2pgm to convert the vobsub files to pgm images using the commandline
vobsub2pgm -v -c 255,0,255,255 inputbasename outputbasename
It appears that I will be able to use the pgm2txt and srttool programs to make these images into a .srt file. gocr needs to be installed for pgm2txt to work and apparently will ask for unrecognized characters. Also, apparently, it's a good idea to run ispell on the output files of pgm2txt.
This info was garnered from this page which concerns DVD ripping and transcoding with Linux.
(EDIT 2005-06-28) I got the subtitles made using the command lines
vobsub2pgm -v -c 255,0,255,255 mysubtitles filmname pgm2txt filmname srttool -s -w < filmname.srtx > filmname.srt
The pgm2txt step required a good bit of user interaction, but not too much.
So, I got things wrong last time. You need to do two passes to get the film looking good. Here are the command lines I used to put subtitles on a film and have it look as good as possible. (Note that the only difference is the number of the vpass.) (Also note that there's an -endpos flag here, because I was doing testing before I committed to doing the entire film. Unfortunately, also, the final size of the film can't be estimated extremely well, so you have to do it once and then adjust the vbitrate accordingly.)
mencoder -sub Hours-eo-utf8.srt -utf8 -font ./arial.ttf -endpos 2:00 \ -oac copy -ovc lavc -lavcopts vcodec=mpeg4:vbitrate=800:vhq:vpass=1 \ -o Hours-subtitoloj.avi Hours.avi mencoder -sub Hours-eo-utf8.srt -utf8 -font ./arial.ttf -endpos 2:00 \ -oac copy -ovc lavc -lavcopts vcodec=mpeg4:vbitrate=800:vhq:vpass=2 \ -o Hours-subtitoloj.avi Hours.avi
Oh yeah, did I mention I'm also adding subtitles here?
Here's the command line I used to re-encode a movie which had been broken into two files. (EDIT: SEE 2005-06-23)
mencoder -oac copy -ovc lavc -lavcopts vcodec=mpeg4:vbitrate=400 -o output.avi file1.vob file2.vob
I want to use my old laptop as a music server, but there seem to be some issues. First, when the screen blanks, the ssh connections drop. This may be controlled by "powersave" and by choosing "presentation" mode in YaST, this doesn't seem to happen. Unfortunately, if I do that, the machine eventually gets really hot and the fan kicks on. I'm looking for a quiet machine, so that's not so great. At the moment, I have it sitting open, which also isn't perfect, but it's not too bad.
I was having a difficult time trying to access a USB drive on my Dell Inspiron 7000 laptop on which I had installed SuSE 9.3. Eventually, the big issue was forcing the load of the "uhci_hcd" module, which I did by hand with modprobe uhci_hcd. Mounting the drive and some other operations are incredibly slow, but mostly everything seems to be functioning well.
Unfortunately, the Dell machine goes into a power saving mode when one closes the lid, so the ssh connections to it seem to die. It's been suggested that other net connections are fine. If that's the case, then I can telnet to the machine, and that'd be fine.
Okay, so I thought it was going to be a simple matter of just putting the DVD burner in for the CDR drive and just creating a bigger ISO. I sure was wrong. First, I had to get the new version of cdrecord-ProDVD from http://cdrecord.berlios.de/old/private/cdrecord.html. This also involves downloading cdrecord-wrapper.sh which includes a security key which is needed in order to write DVDs. Once all that was done, then cdrecord-ProDVD was set up properly.
Second, I had to set up the ISO image correctly. I don't know which one of these made the difference, but I used the "-J" flag to make a Joliet image and "-r" to use the RockRidge extensions with id translation to "good values." After I did that, the DVD was readable on a Mac. I'm not sure if it was readable on a Windows machine, but I'm burning another one now and I'll find out. Ah, yes, it worked!
I was trying to get Mozilla Thunderbird to display dates in the ISO 8601 (my preferred). It turns out that you have to specify the correct LC_TIME value for the locale you want to choose. After a frustrating search, it turns out that the correct value for this is "en_DK" or "en_DK.utf8". So, now using the line LC_TIME=en_DK.utf8 thunderbird, thunderbird comes up with ISO 8601 date . Yay!
Yow! This morning I woke up to find this error message when I rebooted my computer. I remembered that I had done an YOU online update before my last reboot and it had patched the kernel. I *think* that the kernel patch pushed the kernel out of the first 8G of the disk (I had the disk set up as one big partition (well, also with a swap partition) and that was what caused the problem. I have no way to verify this.
So, after booting my other laptop, I searched and figured out what to do. Here are the steps:
Note that this did *not* occur on my machine at school, thereby confirming my suspicion of the cause above, but that also means that the problem may occur on my school machine at some point.
My roommate ripped from a DVD today using the command line
mencoder dvd://7 -o outputname -oac copy -ovc lavc
So, I've been annoyed for a while that when I accidentally hit the [space] bar while holding down the [shift] key, I wouldn't get a space out of the computer. I fixed this by changing a line in /usr/X11R6/lib/X11/etc/xmodmap.std from
keycode 65 = spaceto
keycode 65 = space spaceI can't imagine why this wasn't done already.
WAIT! This did NOT fix the problem. The issue is with mlterm, not with the xmodmap. Rar! And after adding the line Shift+space=" " in /etc/X11/mlterm/key it still doesn't work. Rar! I wonder how to fix this.
Okay, so it took a while to get the Lexmark Z25 color inkjet to work on my SuSE 9.1 machine, but it finally does. Here are the steps that were necessary to get it working.
I had a problem with the hole for my USB cable being too small for the cable itself, and I had to wipe off the color cartridge before the printer would print any blue, but I've got a color inkjet printer now. Neat! And, it will print on REALLY heavy paper.
Okay, so I've wanted to know how to do this for a REALLY long time. here's the trick. In the preamble of the LaTeX file do the following
\DeclareFontShape{OT1}{cmr}{m}{n}% {<5><6><7><8><9><10><12>gen*cmr% <10.95>cmr10% <14.4>cmr12% <17.28><20.74><24.88->cmr17}{} \DeclareFontShape{OT1}{cmr}{bx}{n} {<5><6><7><8><9>gen*cmbx% <10><10.95>cmbx10% <12><14.4><17.28><20.74><24.88->cmbx12% }{} \newcommand{\HUGE}{\fontsize{45}{45}\selectfont}and now use size \HUGE to set your gigantic text. I don't know what the two different sizes after \fontsize are, but at least one is the point size. The "-" after "24.88" is the change that allows larger font sizes to be created. I needed the "bx" version because the large fonts I was creating were in boldface. Also, these lines came out of /usr/share/texmf/tex/latex/base/ot1cmr.fd. If you want to change other font types you'll have to pull the lines from the appropriate files.
I wrote a little script to convert m4a files in the current directory into wav files.
#!/bin/sh # # pooters # nodog # 2004-04-13 # # mym4a2wav # converts all m4a files in the local directory to wav files for M4AFILE in `ls *.m4a`; do echo ; echo --------------------------------------------------------------------- WAVFILE=${M4AFILE/m4a/wav} if mplayer -ao pcm -aofile $WAVFILE $M4AFILE; then echo mplayer -ao pcm -aofile $WAVFILE $M4AFILE; else echo There was a problem with $M4AFILE !!!!; exit 1; fi echo --------------------------------------------------------------------- echo ; done
In order to make an iso from a CD, just put the CD into the drive, make sure it doesn't do some crazy automount, then dd if=/dev/cdrom of=cd.iso where /dev/cdrom is the device for your cdrom. Yahoo!
So, I want a script that'll set up a new account well for me. For this, I guess lighthill will be considered the source.
myR=lighthill.ece.utexas.edu cd ~ rm .bashrc rm .profile touch .profile mkdir offal scp -r $myR:.bashrc $myR:.mailaliases $myR:.muttrc \ $myR:.screenrc $myR:.vimrc $myR:bin .
Everything went well with the default upgrade except I didn't have a keyboard when I first rebooted. Subsequent reboots have had no keyboard problems.
I did lose my USB mouse's scrollwheel (my scrollwheel just didn't work), but I eventually got it back by adding these lines to /etc/X11/XF86Config...
# Combined mice. Section "InputDevice" Identifier "AllMice" Driver "mouse" Option "Device" "/dev/input/mice" Option "Protocol" "ExplorerPS/2" Option "Buttons" "5" Option "ZAxisMapping" "4 5" EndSection Section "ServerLayout" Identifier "Layout[all]" InputDevice "Keyboard[0]" "CoreKeyboard" InputDevice "AllMice" "CorePointer" Option "Clone" "off" Option "Xinerama" "off" Screen "Screen[0]" EndSection(This also fixed a USB hotplugging issue that I had by combining all mice.)
I also got wireless working today. That was pretty neato. It's not nearly as solid as I would like it, but it works mostly. It's the ipw2100 driver which is considered beta at this point. It was pretty easy to install using YaST, but it didn't funcition completely right away. It hung on boot at least once and I can't tell if I had to press Fn-F5 (IBM key sequence) to get it to work or not once (at this point I'm not pressing Fn-F5 on boot at it's working).
With the wire plugged in and wireless starting on boot, I'm grabbing two IP addresses when I boot. Luckily it's not really a problem, because dhcpcd won't reset the default route once that's been established, and the wire connection comes up before the wireless. (I did a quick speed test and I'm getting 575kB/s through wireless and around 6.8MB/s through the wire.)
Okay, so I got ahold of SuSE 9.1 and installed it on a test machine with no real problem. There was a minor problem, but I'll get to that later.
Then, I went to put it on my main machine and I couldn't get the network card to work. It would configure correctly in YaST and the driver would show up on an lsmod, but the device didn't show up in ifconfig and I certainly wasn't getting any network traffic through. I saw errors like "waiting for mandatory device" and "network unreachable."
I eventually tried booting in Failsafe mode, and surprizingly, the network card worked! So, it appeared that some kernel parameter was keepin' me down. After messing with trying all kinds of different kernel parameters, the amazing discovery was that the "vga=0x31a" was the culprit! I changed over to "vga=normal" and the machine and network come up just fine. Weird!
Over the last few days, I've messed with my mouse settings on bleu. Several important changes brought some good results. First, I was able to access the USB mouse (including the scroll wheel) by using the following section:
# Use this one for kernel 2.4 and a USB mouse. Section "InputDevice" Identifier "USBMouse" Driver "mouse" Option "Protocol" "ExplorerPS/2" Option "Device" "/dev/usbmouse" Option "ZAxisMapping" "4 5" EndSectionAnd by making the "SendCoreEvents" change to the ServerLayout section, the USB mouse, the trackpoint, and the touchpad all work together.
Section "ServerLayout" Identifier "Layout[all]" InputDevice "Keyboard[0]" "CoreKeyboard" InputDevice "Mouse[1]" "CorePointer" InputDevice "USBMouse" "SendCoreEvents" Option "Clone" "off" Option "Xinerama" "off" Screen "Screen[0]" EndSection
I've also been trying to get DVDs playing on my new laptop, but playback's been jumpy with mplayer. I started going through all the video drivers (mplayer -vo help) and audio drivers (mplayer -ao help), but that was not the trick. The trick was turning on DMA to the DVD device. I eventually did this with
hdparm -X -d1 /dev/hdcThough it seems like that should've been happening already. I looked in YaST and it says that dma is turned on to /dev/hdc, so I'm not sure what's up. I'll probably need to check this again after a reboot.
Aha. After a reboot, the DMA to the DVD disappeared, so I used YaST to turn it on and now it's part of the /etc/sysconfig/hardware config file.
And dangit if SuSE doesn't always mess with some that I'm used to looking at. Apparently they changed the sort options on ls to something I didn't like, so I had to add the -v option to get back to sorting a way I was used to. Rar!
Yay! I got a new laptop on Monday (2004-05-03) and I've been trying to setup SuSE 9.0 on it since I got it. I've obviously been lax about documenting what I was doing, so I'm going to try to catch up on the story here.
2004-05-03
I got the little beast from a seller on ebay. It seemed to be in good
shape, so I cranked it up. Uncharacteristically, I wanted to use acrobat
reader 6.0, so I figured that I'd just go ahead and boot into Windoze XP and
load acrobat. Well, the unpacking of Windoze onto the harddrive failed. It
hung and wouldn't continue. I hoped this was a failing of Microsoft and not
of IBM, so I just said, "screw Windoze" and started loading SuSE Linux
9.0.
Unfortunately, I ran into big problems there, too. The eventual fix seems to be to choose Reiserfs for the boot filesystem. I like JFS, so I was choosing that. Unfortunately, when I did an Online Update in YaST and loaded a newer kernel (SuSE rev), the machine couldn't reboot. I was getting the GRUB error "Error 2: Bad file or directory type." The kernel updates overwrite the old version of the kernel, so there was no recovering. Yar!
I also ran into a weird problem of the machine slowing way down and keventd taking up lots of memory. That hasn't happened since I went to Reiserfs and have the new kernel update loaded. At one point I notice it started when I hit both the <Fn> and <Ctrl> keys at the same time, so I was worried that it had something to do with the automatic functions built into the IBM keyboard.
2004-05-04
So I got the machine up and running. I decided to run it at 1280x1024 because
that's what I run all my other machines at, so that helps with configuration.
It took a while for me to figure out the sound. I couldn't hear anything even
though the sound card seemed to be configured correctly. Eventually, I
figured out that the sound was just turned all the way down and the volume
buttons above the keyboard were working just fine.
I found some RPMs for lame at pbone and used some other RPMs that I had for mplayer and was able to watch a DVD, though it was jumpy. There seems to be lots of configuration that one can do to mplayer, so I'll look into that soon.
I was able to copy over most of my files using my rsync script. There were some niggling details to that, but they were mostly obvious. The least obvious amongst them was setting the RSYNC_RSH variable to ssh. That took a minute to figure out.
2004-05-05
I got the printer working with a direct connection today, but that wasn't too
hard after I figured out I was missing the YaST printing modules.
LEFT TODO
Later entry: Well, I got wireless working with a PCMCIA card. The two big stumbling blocks were: 1-I needed to use YaST to configure the card (while I was trying to do this, YaST pointed out that the wireless-tools package wasn't installed, so I installed it) and 2-I had to type
route add default gw 192.168.1.1for some reason after removing the ethernet cable and plugging in the wireless card. I would've thought that the wireless script would handle that somehow, but it didn't yet. I'll dig.
The simplest solution is to just do an ifdown eth0 before I plug in the wireless card. That turns off the dhcpcd process and allows the new hotplug of the wireless card to cause a new dhcpcd process.
Okay, so I went back and tried to recompile some old C++ programs. None of them would work, so I wasn't exactly sure what to do. Here's an example of the errors that I was getting...
g++ `wx-config --cflags` -O2 -fno-exceptions -c File.cpp In file included from File.cpp:12: File.h:32: error: 'string' is used as a type, but is not defined as a type. File.h:35: error: parse error before `&' token File.h:38: error: parse error before `&' token File.h:39: error: parse error before `)' token File.h:40: error: semicolon missing after declaration of `File' File.h:40: error: two or more data types in declaration of `openFileIn' File.h:44: error: parse error before `private' File.cpp:22: error: parse error before `&' token File.cpp:28: error: parse error before `&' token File.cpp: In member function `void File::setName(...)': File.cpp:29: error: `fileName' undeclared (first use this function) File.cpp:29: error: (Each undeclared identifier is reported only once for each function it appears in.) File.cpp:29: error: `nameValue' undeclared (first use this function) File.cpp: At global scope: File.cpp:33: error: no `void File::openFileIn()' member function declared in class `File' File.cpp: In member function `void File::openFileIn()': File.cpp:34: error: `ios' undeclared (first use this function) File.cpp:34: error: parse error before `::' token File.cpp:36: error: `cerr' undeclared (first use this function) File.cpp:36: error: `endl' undeclared (first use this function) File.cpp: At global scope: File.cpp:42: error: no `void File::openFileOut()' member function declared in class `File' File.cpp: In member function `void File::openFileOut()': File.cpp:43: error: parse error before `::' token File.cpp: At global scope: File.cpp:51: error: no `void File::close()' member function declared in class ` File' File.cpp:57: error: no `std::fstream& File::getStream()' member function declared in class `File' make: *** [File.o] Error 1
It turns out that the previous compiler (g++ 2.9x) would let me get away with not specifying the standard namespace but the new compiler (g++ 3.3.1) won't. So I needed to make sure that my iostream includes looked like this
#include <iostream> using namespace std;
Now I recompile and things work great. Yay!
So, since BitchX isn't kind enough to let me choose when I do and don't want to use foul language and say offensive things, I've decided to use a different IRC client, irssi. Most of how I wanted to set things up was pretty simple except for a change from irssi's default behavior of openning a new window and starting a "query" whenever either I /msg'd someone or someone /msg'd me.
Eventually, I figured out the set of commands that I had to issue to make the windows work the way I wanted. Those commands are executed in the status window.
/window level -msgs -dccmsgs /set autocreate_query_level none /set autocreate_own_query off
Those commands seemed to do the trick. The one that was the hard one to find was the first. I didn't understand how irssi was handling "window levels" well enough to think to look for it.
So, the Online Update told me to that I'd need to reinstall my nvidia X driver in order to continue, and just as last time I was told this, my screen split in too when I restarted X. So I stopped the display manager (sudo /etc/rc.d/xdm stop), went and got the newest nvidia Linux driver from here, and tried to install it (sudo sh NVIDIA-Linux-x86-1.0-5336-pkg1.run). Of course it couldn't be installed and I pondered for a bit. After exploring, I realized that the running kernel (uname -r) and the kernel that was installed to run (ls -l /boot) weren't the same. I rebooted, retried the above installation command, and the nvidia driver went in with no problem.
The SuSE install README suggested that I then use SaX2 (sax2 -m 0=nvidia) to configure X, but that was unnecessary. If I'd been installing from scratch, it would've helped. I did install the NoLogo option (Option "NoLogo") into /etc/X11/XF86Config in the "Device" section.
I wanted to use a USB drive on lighthill but hotplugging for USB wasn't working. I beat my head against it for a long time. The first thing I did was turn on the USB device on the motherboard. The last thing that I did was to change the value of HOTPLUG_START_USB from "no" to "yes" in /etc/sysconfig/hotplug. If I'd only done those two things, I would've gotten it working immediately, but I had to go off searching to find out that was what I needed to do. Yar!
In order to print Esperanto (or unicode in general) from mozilla/firebird, a program called wprint can be used on the postscript which is output from mozilla/firebird. It messes up the spacing, but there doesn't seem to be anything you can do about that.
I tried to install wprint as just myself into /usr/local but wprint expects it's configuration file to be in /etc. Bummer. I had to install as root (sudo make install).
The configuration file (/etc/wprint.conf) required that I knew the path to truetype fonts on the system. Here's a snippet of the configuration file that shows what I changed. The times.ttf truetype font worked best for me, even better than the cyberbit.ttf font that the program recommends.
fontpath:/usr/X11R6/lib/X11/fonts/truetype/ uni:UTF-8:times.ttf:wprintout.ps
The process to print unicode from firebird now is to first print to a file, then wprint that file. That produces a new file which is then sent to the local windows printer using smbclient. I know it's complicated, so it's good that I don't have to print unicode too often.
The problem wasn't as simple as I thought. It turns out that I can't simply convert all the file and directory names to lowercase because *some* of them are called by their uppercase names within the program. AND Matlab (my reason for needing to do this at all) won't use a link as a directory name. Geez!
The final answer was to mv $NameWithUppers $NameOnlyLower then ln -s $NameOnlyLower $NameWithUpper for every file in a directory during a depth first search. The final script looked like
cd "$1" for NameWithUppers in *; do NameOnlyLower="`echo $NameWithUppers | tr '[A-Z]' '[a-z]'`" if [ "$NameWithUppers" != "$NameOnlyLower" ]; then echo U:"$NameWithUppers" L:"$NameOnlyLower"; if mv "$NameWithUppers" "$NameOnlyLower"; then echo mv "$NameWithUppers" "$NameOnlyLower"; else exit 1; fi if ln -s "$NameOnlyLower" "$NameWithUppers"; then echo ln -s "$NameOnlyLower" "$NameWithUppers"; else exit 1; fi fi doneand the find command line was
find . -depth -type d -exec ./mylowerlink {} \;
I have two computers at work, lighthill and helmholtz, which are connected on a linksys hub/router. I filled up helmholtz, the ftp server, and needed to move some files to lighthill and then share them over NFS back to helmholtz. I changed the appropriate files (/etc/exports and /etc/fstab) by hand and tried to mount -a on helmholtz, but I kept getting the error "RPC: Program not registered." Then I finally realized that I hadn't started the NFS server on lighthill. I went into yast2 and used their interface to start the server on lighthill and then everything went as planned.
I needed to change all of the directory names and file names of a directory tree to lowercase. After trying several things, I eventually decided upon using find and a script which first cd-ed to the directory then used the packaged script lowercase_filenames to change the names to lowercase. The command line looked like...
find . -depth -type d -exec lowerit {} \;and lowerit contained
cd $1 lowercase_filenamesThis seemed to be the only way to get find to execute "two" commands on what it found.
There are some problems with this method, mostly caused by the lowercase_filenames script which tries to handle problem cases and starts making new directories. Basically, this script works the first time, but will munge things the second time. (I just realized that the problem is actually caused by having a directory which would become named something that already exists.) If I change the lowerit script to be
cd $1 for f in *; do mv $f `echo $f | tr '[A-Z]' '[a-z]'` donethen I don't get the same problems with running it more than once and the safeguards in mv keep me from trying to overwrite a file with itself or moving a directory into a itself as a subdirectory. (This isn't an optimal solution, but it is a solution.)
So I'm interested in not only putting unicode on my webpages, but I'm also interested in printing unicode text files as a regular text file. So far, I've run across two solutions, cedilla (supposedly a replacement for a2ps) and uniprint. I'm going to look into them and see which one's best for me.
Well, at the moment, neither of them look very good. Bah. I'd rather use cedilla since it's part of GNU, but it doesn't seem to handle tabs correctly. uniprint needs a font to use, but then I guess enscript (the program I currently use for text printing) does, too. enscript does seem to handle tabs correctly.
I decided on uniprint because it has more fonts (TTF) and seems to handle spacing a little better (not as well as enscript, though). It also allows pipelining. Unfortunately, it's part of yudit and that package is 10MB (cedilla's is 1MB).
So, I've been learning Esperanto and now I want to be able to type and publish web pages in it while using Linux. This has taken a good bit o' figurin', but I think I've finally gotten it. The solution is Unicode and specificially the UTF-8 encoding. Most of this info is culled from two sources,
First, it is best to change one's locale to a UTF-8 locale (current is
found by typing locale and possibities are found by typeing
locale -a. For those with English as a primary language, then the
appropriate addition to .profile (for bash) is
export LANG=en_US.UTF-8
Though this may be different for users of other Linux distros (mine=SuSE 9.0)
Next, one must use a UTF-8 capable terminal program. The only two that I was able to find were the newest xterm and mlterm. If your locale is setup correctly, then mlterm is ready to go. You'll have to specify a font for xterm. The best seem to be the -misc-fixed-*-iso10646-1 fonts.
Now, you should be set to display UTF-8 characters well. The problems is now typing them in. My editor choice is vim, so I can just use a keymap to be able to type in characters easily. Within vim, I just :set keymap=esperanto_utf-8 and now I can use any one of several established systems for typing Esperanto into text terminals and the correct UTF-8 code points are entered for me. Yay!
In order to display UTF-8 on a web page, one must include a
META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=UTF-8"HTML directive in the HEAD section of the page.
And that should be it. Good luck.
So, it was quite a fight to get SuSE Linux on my labmate Nick's computer. He has a K7N420 Pro motherboard from MSI and it uses both nVidia nForce2 integrated peripherals and an nVidia GeForce2 Integrated GPU video card.
The steps which finally seemed to make a difference were...
After that, things worked. This information was found from the README file that came with NVIDIA-Linux-x86-1.0-4496-pkg2.run (in its /usr/share/doc/ directory). That's a large and extensive file about the drivers that they've created.
So a friend sold me a 13G IDE drive for $15. This means that I now have a little more room on the PIII-500 that the school gave me. Unfortunately, I can't boot off an LVM, so I'm just going to throw the 13G at my home directory (with a 512M nod to swap space). I know that I *should* make a tiny boot partition and then LVM the rest of the original 6G together with the 13G, but since I already had so much stuff installed on my new machine already, I'm not gonna do that.
I'm going to finish moving everything over to the new lighthill and take the old lighthill home to give to Nina and her house.
I'm now trying to burn a CD from the rar files I downloaded from a bit
torrent link. Here's the command line I'm trying...
sudo cdrdao write --device 3,6,0 --speed 12 susel90c.cue
This worked. Yay!
I put in the SuSE 9.0 Pro CDs and tried to do an upgrade. This involved going to "Installation" and then YaST recognized that there was already a system on the machine. It then went into the "update" option. This was working fine until... the installation wouldn't recognize CD3. Yargh! I ended up "aborting" so I don't know what the system is going to do now.
Well... the machine came up, so we'll see. Seems to be going well so far. I still need to update some packages, but I can do that with "System Update" in yast2 once I get a new copy of disk 3.
The powers that be saw fit to give me an old Dell Optiplex GX110 (PIII-500MHz) PC at the lab, so I'm installing the new SuSE 9.0 onto it. Unfortunately, though the CPU is speedier, it has less memory and only a 6.4G HD. I'm not sure what I think of the situation, but I'll give it a go.
Hmmm... Though the installation of SuSE 9.0 went well, I think that the machine isn't any faster than my PII-333 was because it has a super-zippy SCSI drive in it.
I wonder if I can scrape up enough cash to get myself a new laptop... Hmmm... Then I could take my home desktop, throw it at the lab, and then keep my laptop at home. Hmmm...
Argh... so after an "Online Update" which upgraded the kernel (and kernel source, apparently), the NVIDIA video drivers died (wouldn't start). Turns out that this was an issue between the kernel version and the version that the NVIDIA drivers were compiled for (Athlon).
In order to get things working again, I had to re-install the NVIDIA drivers. In order to do that, the installation program needed to recompile the kernel interface. In order to do that, I needed to do something listed in /usr/include/linux/version.h...
- Make sure that the symbolic link /lib/modules/`uname -r`/build exists and points to the matching kernel source directory - Now copy /boot/vmlinuz.version.h to /lib/modules/`uname -r`/build/include/linux/version.hOnce I did the second one, I was able to re-install the NVIDIA video drivers. Yay!
I wanted to be able to work in /usr/local as a trusted user...
sudo chgrp trusted /usr/local/* sudo chmod g+w /usr/local/* sudo chgrp -R trusted /usr/local/man sudo chmod -R g+w /usr/local/man
Wow, it's been a royal pain in the tookus to get Csound running on SuSE8.2.
So, the place to go is Bath University and get the file Csound.tar.gz from the newest directory.
In order to get this to compile, I needed to install the "gtk2-devel", "fltk" and "fltk-devel" packages.
In Makefile I had to
- add "-I/usr/X11R6/include" to the CCFLAGS and CXXFLAGS variables
- change references to "-malign-loops" and "-malign-jumps" to "-falign-loops"
and "-falign-jumps" respectively
- uncomment the line that sets the variable FLTKLIB to the value "-lfltk
-lpthread -lg++" so it can load dynamic libraries instead of static libraries
(the default setting of the FLTKLIB variable in the makefile)
(NOTE: if you want to use the static libararies, you'll need to change the
FLTKLIB variable to point to "/usr/X11R6/lib/libfltk.a" instead of
"/usr/local/lib/libfltk.a")
I also had to go into the source file "widgets.cpp" and change the reference to "strstream.h" to "strstream" and "fstream.h" to "fstream" (don't ask me what's up on this one).
I had to make several symbolic links in /usr/local/lib...
ln -s /usr/lib/libg++.so.2.7.2 libg++.so ln -s /usr/lib/libg++.so.2.7.2 libg++.so.0 ln -s /usr/lib/libtcl8.4.so libtcl.so ln -s /usr/lib/libtcl8.4.so libtcl.so.0 ln -s /usr/lib/libtk8.4.so libtk.so ln -s /usr/lib/libtk9.4.so libtk.so.0
(which will obviously need to change when the versions of g++, tcl, and tk change.)
And run sudo ldconfig again.
Comiled with "make" and installed with "make install" (because I'm a trusted user on my machine).
Yay! It works!
There are some suggestions that I should have just gotten the binaries from http://www.csounds.com/istvan but, unfortunately, that section of csounds.com seems to be down at the moment.
Since I switched to SuSE, there've been some configuration issues with
vim.
1. It would immediately move to the last place edited in a file.
2. It wouldn't wrap on backspace while inserting.
3. It paused quite a bit (especially at the beginning of cursor
movement).
I solved (1) by editing the /etc/vimrc and removing the `if has("autocmd")' lines. (2) should be handled by the "whichwrap" option. I'm still looking at a way to fix (3), though the "swapsync" option looks promising. I changed it to "sync" from "fsync" as suggested in the helpfile. I may change it to not be set at all if the pauses keep happening.
NOTE: "swapsync=sync" seems to be behaving very nicely.
19jun03 NOTE: (2) is handled by "backspace" option
So, suddenly, for no apparent reason, my Dell Inspiron 7000 laptop (gouda) decided to not turn the backlight on again after returning from a blank screensaver. Dunno why. I ended up resetting the BIOS options so that the computer doesn't go into STANDBY or SUSPEND, but it's still occurred since then.
The quickfix seemed like it was going to be pushing the "laptop is closed" button for about 20 seconds and then waiting for the SUSPEND to cycle through. A better option seems to be cycling through the CRT/LCD modes with Fn-F8. I don't know what the long term solution will be.
So, a manga that I read, Hikaru no Go, requires Bit Torrent to download. I can't find any RPMs for SuSE, so I had to install by hand...
tar xzvf BitTorrent-3.2.1b.tar.gz cd BitTorrent-3.2.1b sudo chgrp trusted /usr/local/bin sudo chmod 775 /usr/local/bin chmod +x bt*.py mv bt*.py /usr/local/bin sudo mv BitTorrent /usr/lib/python/site-packagesI'll use mozilla's "helper" preferences to use btdownloadgui.py to download bit torrent links. Yay!
This is intended as a list of things that I need to do for a new shell account (not a graphics workstation).
Looks like my motherboard's chipset (nForce2) isn't supported in linux kernel 2.4.20. It is supported in 2.4.21. So, I'm going to try a friend's UltraDMA100 PCI IDE controller tonight and see if that helps with my problems. I am hopeful.
YAY! This did the trick! My pausing/jumpiness/whatever problems seem to be fixed by having a new HD controller. Yay! Now, I can just keep using Matt's controller until the 2.4.21 kernel comes out which has the nForce2 HD driver in it.
I couldn't get mozilla to run. I didn't know why. I would type in mozilla and it would crank a bit, then just quit with no messages. After looking around on the web, someone mentioned that they could run mozilla as root, but not as a user. I tried with sudo mozilla and it came right up (with lots of "generating font" messages, thoough). Yargh! So then, I tried again as a regular user and I got the "create profiles" window. I tried to create a profile and it got stuck when I went to [Finish]. I took a look at my ~/.mozilla directory and it along with several subdirectories were owned by root:root (user:group). I chown-ed them to nodog:users and now mozilla seems to run just fine. What a pain in the Targus!
Also, my system beep (bell, notification boop) seemed to be missing. For whatever reason, my PC Speaker channel on my sound card had been set to 0 volume. Setting this back to 100 (in YaST, of course) fixed the problem.
So, I was trying to get wireless networking to work and it just wouldn't.
I was getting errors like...
eth0: Error -110 reading firmware info. Wildly gue ssing capabilities... eth0: Station identity 0000:0000:0000:0000 eth0: Looks like an Intersil firmware version 0.00 eth0: Intersil firmware earlier than v0.08 - sever al features not supported eth0: Ad-hoc demo mode supported hermes @ IO 0x100: Error -16 issuing command. eth0: failed to read MAC address! orinoco_cs: register_netdev() failedin /var/log/messages
Then I remembered that when I set it up the first time I had found (from a google
search) that a Dell Inspiron 7000 laptop using an orinoco/wavelan card had to
have certain ports excluded in /etc/pcmcia/config.opts. I changed
include port 0x100-0x3af, port 0x3e0-0x4ff, port 0x800-0x80f, port
0x820-0x8ff, port 0xc00-0xcff
to
include port 0x300-0x3af, port 0x3e0-0x4ff, port 0x800-0x80f, port
0x820-0x8ff, port 0xc00-0xcff
and added the line
exclude port 0x100-0x2ff
and then things worked as I expected them to. Yay!
Okay, so I like SuSE a lot and I'm going to install it on both my lab computer, lighthill, and my home laptop, gouda.
First, I plugged in an old 4G disk to back everything up with. I just used the whole thing as one big partition.
sudo cdfisk /dev/sdb sudo mkfd -v -t ext2 -m2 /dev/sdb1 sudo mount /dev/sdb1 /mnt cd /usr sudo cp -pr home /mnt cd local sudo mkdir /mnt/usr sudo mkdir /mnt/usr/local cp -pr BIG backup bin down ears eyes faq hikaru kgscgoban share /mnt/usr/local
So, I went through the install and everything went well (except perhaps the X configuration, which I had to Ctrl-Alt-Backspace out of).
After the install, I needed to
- install all users
- put myself in the sudoers file
- make it possible for me to write into /usr/local as a user
- copy the backup info back onto the machine
- fix the file owner and groups on the stuff I copied back
- get mail working
- get the sound card working (trivial in YaST)
- add "server now.okstate.edu minpoll 10 maxpoll 13
server clock.psu.edu minpoll 10 maxpoll 13" lines to /etc/ntp.conf - get
xntpd put into /etc/init.d/rc5.d (K09xtnpd and S09xntpd link)
After that was done, I needed to work with my dotfiles to get the
environment that I was used.
Simple ones:
- .screenrc
- .muttrc
- .mailaliases
- .project
- .plan
- .vimrc
Trickier ones:
- .Xresources (.Xdefaults is just a link to this)
- .profile (.bashrc is just a link to this)
- .fvwm2rc
In order to use the windowmanager/desktop selector provided with SuSE, I needed to use the provided .xsession and modify my .fvwm2rc so that it starts up the xload, oclock, xbiff, and aterm programs the way I like. This is a better way to do things anyway.
I also needed to run ssh-keygen -t rsa.
gouda would hang during the boot from the SuSE CDROM 1 (yast2 wouldn't start). After looking online for a long time, I found these two pages ( 1 2 ) which are about installing SuSE 8.0 and 8.1 on a Dell Inspiron 7000 laptop. The big trick was booting in rescue mode and then running hwinfo. That program told me which piece of hardware was failing and I could then put in "hwprobe=-serial" in my boot prompt. Yay!
PCMCIA didn't come up automatically, unfortunately, but that's not totally surprising since I have WEP enabled. Removing WEP didn't fix it immediately. Ugh. For now, I plugged in the wired LAN card that I have and it worked.
I went back and realized that the sound wasn't working on pimento. Urgh. I fiddled with it, very much like I fiddled with getting the network, but to no avail. Whatever I try to play repeats just the first second over and over. Oh well.
E: Archive directory /var/cache/apt/archives/partial is missing. E: Tried to dequeue a fetching object ... E: Tried to dequeue a fetching object update available list script returned error exit status 1.and I need to recreate the above directories.
Okay, so I'm up for putting SuSE 8.2 on pimento. Hope it goes well.
First problem is that I needed to make boot disks. I made them from the
CDs with command lines...
dd if=offal/bootdisk of=/dev/fd0u1440
dd if=offal/modules1 of=/dev/fd0u1440
dd if=offal/modules2 of=/dev/fd0u1440
Then things got stuck when I inserted the "modules2" disk. After searching on the SuSE site, [note: this link was changed from referencing www.suse.com on 2008-12-16 at Novell's request] I found that the "modules2" disk image is corrupt, so I had to download a couple of new images of "modules1" and "modules2". Other than that, things are going well...
Changes to the suggested config...
Divided the 57.2G drive into a 56.2G root and a 1G swap (this is the first
time that I've had a "single" partition for linux in a long while).
Lots of software changes...
NOTE: The jumpiness problem still exists and is affecting the
installation, which is therefore slowed down. renice isn't on the
installation disk, so I'll have to wait until the install is done before I
can issue the fix from above.
Okay... problems with the net drivers (well, duh). Following instructions on the SuSE site and instructions on the nVidia site, I never got it working... here's how.
I downloaded the source RPM from the nVidia site. Then, I had to make sure I had the the following pacakges: make, gcc, binutils, glibc-devel, kernel-source. Through some trial and error, I eventually discovered that I needed to copy /boot/vmlinuz.version.h to /lib/modules/2.4.20-4GB-athlon/build/include/linux/version.h because this wasn't done correctly. Then I executed
rpm --rebuild NVIDIA_nforce-1.0-0248.src.rpm
which put a new RPM in /usr/src/packages/RPMS/i386. So, then I
rpm -i NVIDIA_nforce-1.0-0248.i386.rpm
OKAY... FORGET IT!
I went and bought a RealTek 10/100 card so I don't have to mess with this
stinkin' nvnet thing. It worked immediately as a Tulip card. Yay!
So, I'm going to try the tech's suggestion and put a swap file on the
"extra" windoze disk I have in pimento. First I mounted the drive in the file
system...
sudo mkdir /windoze
sudo mount -t msdos /dev/sda1 /windoze
I then added this to /etc/fstab. Now, I'm creating a 768M file using...
sudo dd if=/dev/zero of=/windoze/linux.swp bs=1024
count=786432
then enabling it as a swap file with...
sudo mkswap /windoze/linux.swp
Finally, I'm enabling swapping using this file by adding a line to
/etc/fstab...
/windoze/linux.swp none swap sw 0 0
and running...
sudo swapon -a
I also temporarily disabled swapping on /dev/hda2 by executing...
sudo swapoff /dev/hda2
I guess it's time for a test of the jumpiness...
Bluh. I'm still getting the same behaviour. What a pain in the axle
grease.
So now I'm reconfiguring the kernel to mostly use modules instead of compiling everything into it (my standard)...
cd /usr/src/linux sudo make clean sudo make mrproper sudo cp ~/adm/pimento/usr/src/linux/config .config sudo make menuconfig sudo make dep sudo make sudo make bzImage sudo make modules sudo make modules_install sudo make install cd /usr/local/down/nvidia/nforce sudo make install sudo reboot
I eventually had to go back and redo this process from a completely clean install of the kernel source, but that worked just fine.
More testing of the jumpiness...
First test... no jumpiness... need longer test...
Okay... needed to get X11Forwarding working on pimento, so changed the line
in /etc/ssh/sshd_config from "X11Forwarding no" to "X11Forwarding yes".
Second test... still jumpiness. Yuck.
HEY! I think I figured it out! From some weird mailing list info on ksoftirq I reniced the ksoftirqd process to priority 0 and things seem much happier! Yay! I found this in the NAPI_HOWTO, but I don't know where that document is usually supposed to reside. And, I don't know why I would need to raise the priority level, but I did.
ksoftirqd is in the kernel, so the process isn't actually "started" by anything. I wrote a little script (below) to renice the priority when I start seeing troubles.
#!/bin/bash # # pooters # nodog # 09may03 # # mypimentofix # fixes the ksoftirqd_CPU0 issue on pimento by renicing that process to 0 MYPROCESSNAME=ksoftirqd_CPU0 if MYPROCESSPID=pidof $MYPROCESSNAME` ; then echo $MYPROCESSNAME has PID = $MYPROCESSPID; sudo renice -0 $MYPROCESSPID; else echo There is no process with the name $MYPROCESSNAME; fi
Turns out that the new SuSE has XFree86 4.3.0, so I might switch over. I've been looking for a reason to do this for a long time, but Ive been hesitant because of learning curve issues. My friend, Matt Hudson, claims that he's installed SuSE in less than an hour, so I may switch.
The machine is currently at Laboratory Computers being fixed on the $35+parts agreement.
Now it's back. Got a new processor, heat sink, and fan. The tech, Tony, told me that the processor was burned up all over which is inconsistent with overclocking. He thought that the machine had been suffering for a while. I also got a new video card from them, 'cause my nVidia Riva TNT2 wasn't working well. I purchased an nVidia GeForce4 MX440 64M. He says that it works with Linux, so that's good enough by me for now.
The tech watched me make the jumpiness appear and his suggestion was that it was a disk caching problem. He said that if I was having trouble like that while doing that heavy processing, I need to either have my swap on another drive, or have two drives raided. Interesting.
So, for now, I'm starting to reload packages and whatnot... X first, of course, since sound and net are both already working. Okay, so X doesn't support this new card until version 4.3.0, which is out, but isn't yet in Debian's testing distro.
Before I start on the Linux reinstall process, I'd like to try changing out the video card once again. Now, I have a much better idea about what's up with the PCI Slot IRQs, so I hope that the "old" card that I had lying around will work (I hope it fixes the jumpiness/halting problem, too!).
Trying the PCI Diamond Fire 1000 GL (broken nVidia TNT2 64M card currently
being used)...
pimento sticks on boot.
Trying to use the TNT2 card to assign an IRQ to Slot 1...
now pimento gets stuck on boot in the same place as when the Diamond card was
in.
Now removed Adaptec SCSI controller and trying Diamond card again...
pimento seems to boot, but I get no video.
Back to TNT2 card... set PCI slot 1/4 to IRQ 12.
Trying Diamond card again... different sticking place, no video.
Trying an old STB Velocity 3D card... boots correctly... looks good so far...
(video is much cleaner!)
BUT, most importantly, this does not fix my jumpiness problem.
Onward with the reinstall! (leaving the STB in the machine for now, 'cause it
looks better)
Just for the heck of it, I copied /usr/src/kernel-source-2.4.20.tar.bz2 to /mnt/nodog while /dev/sda4 was mount -t msdos /dev/sda4 /mnt.
- Getting the rescue, root, and driver disks (2) for "compact".
- Booting from rescue...
- inserting root...
- choosing language...
- choosing keyboard...
- (had to go backwards in the menu to get to) partitioning the hard disk...
(1G root, 1G swap, rest as /usr)
- activate swap partition (with bad block check, took ~8 min)...
- initialize linux partition (with bad block check, ~8 min )...
- mounted root file system...
- initialized another linux partition (without bad block check (it's
60M!))...
- mounted /usr file system...
- wow... installed kernel and driver modules (from woody CDROM I
downloaded)...
- could've configured the drivers, but there are none that I need right
now...
- configured hostname...
- install the base system (from the woody CDROM again)...
- make system bootable...
- reboot the system... now must configure the Debian system (can revisit this
with /usr/sbin/base-config)...
- local machine time, central time zone, md5 passwords, shadow passwords, set
root password, set nodog user account, removed pcmcia packages, no PPP, don't
scan another CD, no other apt sources, no security updates (yet), ran tasksel
(?!?), ran dselect (wow, this was a bit funky), added mime handler for
"application/*", set locales to be generated (what? chose en_US ISO-8859-1
and en_US.UTF-8 UTF-8), default locale of C, allow ssh2 only, added
sunfire.ece.utexas.edu now.okstate.edu clock.via.net to
/etc/default/ntp-servers and /etc/ntp.conf, mail set to "internet site"
Oops! Forgot to backup my "." files. Will have to recreate those.
First thing that I'd like to do is check and see if I still get jumpiness. Hmmm... everything that I know of that caused it involved translating audio files. Hmmm... do I need net access first to try this out? How about just unzipping the kernel source?
added nodog to /etc/sudoers with visudo
sudo mount -t msdos /dev/sda1 /mnt cd /usr/src sudo cp /mnt/nodog/kernel-s.4 kernel-source-2.4.20.tar.bz2 bzcat kernel-source-2.4.20.tar.bz2 | sudo tar xvf - sudo ln -s kernel-source-2.4.20.tar.bz2 linuxDuring the bzcat-ing, I messed around in another window. Things don't seem quite as fast as they should be, but it seems much less jumpy. I need to try some audio encoding and whatnot, so I need to get net working. I should've just kept that NetGear card that Mike lent me. D'oh!
sudo mount /cdrom
sudo vi /etc/group --- added nodog to staff
cd /cdrom/usr/local
cp -pr * /usr/local --- at this point I notice some jumpiness.
ARGH!!!
cd /usr/local/down/nforce
sudo make install --- Do'h! I'm going to need to make my own kernel to
get these working.
Must get config file on pimento for kernel config. Need mtools available
to nodog for that to happen.
sudo vi /etc/group --- added nodog to floppy
cd ~
mcopy a:config .
cd /usr/src/linux
sudo cp ~/config .config
sudo make config --- pressed ENTER about 1000 times.
sudo make dep --- didn't really notice jumpiness
sudo make bzlilo
sudo make modules
sudo make modules_install
sudo reboot
cd /usr/local/down/nforce
sudo make --- problems... yuck (eventually solved by just
repeating)
sudo make install
added an eth0 line in
/etc/modules
added an alias eth0 nvnet, alias sound-slot-1 i810_audio,
and alias usb-interface usb-ohci lines in
/etc/modutils/aliases
ran sudo update-modules
Networking was not working. At this point, I needed to do some exploring.
Eventually I remembered that I hadn't been able to configure networking
during the installation, and that needed to be set up. After RTFM, I needed
to get the etherconf package in order to set up networking. This also
required the libconfhelper-perl and liblogfile-rotate-perl packages, which I
went and got as well. Eventually, I was able to
sudo dpkg -i liblogfile-rotate-perl_1.04-1_all.deb
sudo dpkg -i libconfhelper-perl_0.12_all.deb
sudo dpkg -i etherconf_1.10_all.deb
and get to the networking
configuration. After typing in the appropriate info, networking is up and
running. Yay!
Setup /etc/hosts and /etc/apt/sources so I could use dselect to switch over to the testing distribution and get myself started back on that path.
Must also get flac or mpg123 packages to try out jumpiness.
dselect upgraded many packages and asked for amny configuration decisions. I'm not listing them all here.
Things went funky somehow and the kernel was crashing... solved with...
cd /usr/src/linux sudo make clean sudo mcopy a:config .config sudo make dep sudo make bzlilo sudo make modules sudo make modules_install sudo reboot cd /usr/local/down/nforce make clean sudo make install sudo reboot
Everything seems ready to test for jumpiness (but I'm going swimming!).
Okay, so I tested for jumpiness. It's still there. Damn. So I googled a bit and found some mailing list stuff about underclocking AMD Athlons to make some linux problems go away (specific to dual processor boards, but I'm at my wit's end here). I underclocked. The jumpiness stayed.
So then, I thought to myself, why not overclock and see if anything good happens. D'oh! I seem to have broken my processor. The machine won't do anything now but blink the three keyboard LEDs and (thanks to Microstar's diagnostic LEDs) stay at the "processor is broken" state. Fudge.
BIOS setupTrying different rescue floppies...So, I can set the 2nd boot device to the SCSI controller and boot there. That way I can boot from the SCSI CDROMs. This means, however, that I'll need to write a LILO MBR onto the SCSI disk, which is currently Windoze only. Not sure if I want to do this, but it may be an option later.
Perhaps try disabling the cache while running the installation.
Trying this... Nope... still gets stuck after the line
Serial driver version 4.27 with HUB-6 MANY_PORTS MULTIPORT SHARE_IRQ enabled
Okay, so now I need to make backups so I can do an install.
- I need to backup ~/BIG and ~/ATTIC on pimento.
- I need to make sure I have a copy of the net and sound drivers for pimento
(these are in /usr/local/down which is part of the "backup stuff"
below).
- I need to backup the X config, kernel config, listing of /usr/local/bin,
/etc, /usr/local/etc, /usr/local(backup stuff), and anything else I can think
of. (My home directory is already backed up.)
Backing up BIG
mkisofs -A BIG -o big.iso -P AndersonMills -p AndersonMills -R
BIG
sudo cdrecord -v -eject -data big.iso
Backing up ATTIC
cd ~/ATTIC/Hikaru-episodes mkisofs -A disk18 -o disk18.iso disk18 && \ mkisofs -A disk19 -o disk19.iso disk19 && \ mkisofs -A disk20 -o disk20.iso disk20 && \ mkisofs -A disk21 -o disk21.iso disk21 && \ mkisofs -A disk22 -o disk22.iso disk22 && \ mkisofs -A disk23 -o disk23.iso disk23 && \ mkisofs -A disk24 -o disk24.iso disk24 sudo cdrecord -v -eject -data disk18.iso sudo cdrecord -v -eject -data disk19.iso sudo cdrecord -v -eject -data disk20.iso sudo cdrecord -v -eject -data disk21.iso sudo cdrecord -v -eject -data disk22.iso sudo cdrecord -v -eject -data disk23.iso sudo cdrecord -v -eject -data disk24.iso cd ~ mkisofs mkisofs -A ATTIC -o attic.iso -P AndersonMills -p AndersonMills -R ATTIC sudo cdrecord -v -eject -data attic.iso
Backing up "backup stuff"
cd /usr/tmp/nodog tar xzvf ~/adm/cdbackup.tgz cd cdbackup/pimento/ sudo mkisofs -A pimentoADM -P AndersonMills -p AndersonMills -o pimentoADM.iso -R -f admin sudo cdrecord -v -eject -data pimentoADM.iso
Now, I'm trying to get back to where I was before I had the NetGear card in pimento.
So, to get back to where I was, I...
This didn't work, so, I need to go back and check what I did when I got networking working the first time. On 27jan03, I got and installed some nVidia nForce drivers for this motherboard. According to the notes there, for every kernel build, after every
sudo make modules sudo make modules_installI also have to go to /usr/local/down/nforce and do a
sudo make installI did this, rebooted, and the network seems to be happy again. Yay!
There seems to be issues with the SCSI interface now, but I'm not sure what's up there. I think that the NetGear card took over the IRQ that my BusLogic 946C SCSI controller was using and now the BusLogic SCSI controller is unhappy. As long and I can burn CDs, I'll be happy for tonight.
Let us test...
Nope. We got problems.
However, when the machine booted this time, there was an fsck on /dev/hda1 and an error was found. pimento booted into maintenence mode. I fsck-ed and reboot-ed.
I immediately went to /dev and executed sudo ./MAKEDEV sda. It worked with no errors (as opposed to running this before). Hopefully this also fixes my "dual booting to Windoze" problem I just discovered earlier today. Lesse... sudo lilo returns no errors. Yay! Further testing shows that Windoze loads just fine now. Yay!
Just because I'm geeking out anyway, I'm going to try removing the i810 NVidia nForce Audio driver from the kernel and see what happens (Why the hell amd I doing this while I'm burning a CD? Oh well... it burned just fine.)...
cd /usr/src/linux sudo make menuconfig sudo make dep sudo make bzlilo sudo reboot(I didn't need to remake the modules since I only removed a driver from the kernel). Argh! No effect, so I put the i810 NVidia nForce Audio driver back into the kernel. I wish I could figure this out!
So, pimento pauses every once in a while during heavy load for up to a few seconds at a time. This behaviour is really annoying! I've been switching pieces in and out of the computer to see if I could fix this.
I'm sticking a NetGear FA310TX Rev-D2 NIC in it to see if the onboard NForce LAN adapter is part of the culprit.
cd /usr/src/linux sudo make menuconfig sudo make dep sudo make bzlilo sudo make modules sudo make modules_install sudo reboot
I ended up having to reboot twice (for no discernable reason), but then the DEC Tulip 10/100 driver worked flawlessly. Unfortunately, the new net card had no effect on the pausing. I even stopped running X and was able to make the machine pause just like before. Argh!
At some point, I'd like to replace the video card which I know is at least partially broken.
- okay, so, I think I've got local small enough
- now, I want to store the MP3s on earth
* need to squish things for AMODA including making better anims
DONE - switched to MNG
* need to go through audition articles, especially ZZ-unknown
- wow the flac codec rocks!
- running out of space on gouda... need more room
- my home directory is huge
- /usr/local is pretty big
- may want to take some unused packages off, too
3 largest dirs on /usr are local, home, and share
I want to look @ local and home
in local --- eyes, ears, dirt, hikaru
in home --- csound, phyd
* must go get "imtheone" stuff from /usr/tmp/nodog/RSYNC/backup on
pimento
DONE
* must put dirs in BIG
DONE
* must fix myrsync to take pimento as repository
* must make myrsync-BIG to sync BIG b/w pimento and lighthill
* also tell where it's pushing to
BIG
* look @ monkey films in /usr/local/eyes/humor
DONE - different kept both
* process garcia pics
DONE
* flipside pics? do I need 'em?
DONE - processed, smaller, kept
Okay... I've been having creeping problems w/the devices on pimento for a while and now it's finally bitten me.
This was eventually cleared up with an fsck on /dev/hda1. This is weird 'cause #1 the symptoms didn't point to this, and #2 it looked like it was getting fsck-ed each boot, but I guess it wasn't.
pimento is being all jumpy when the processor gets heavily loaded. This is not behavior that I've seen in other linux machines, so I'm gonna try to figure this out.
Mike Stone suggested passing either pci=acpi or acpi=off to the kernel on boot.
Initially, acpi=off seemed to correct the problem.
NOTE: The jumpiness shows back up. -- 02may03
Looks like the Linksys router box needed to be reboot, too. After rebooting it, the network was significantly less jumpy.
- must recompile kernel to include wireless support
- in /usr/src/linux
sudo make menuconfig sudo make dep sudo make bzlilo sudo make modules sudo make modules_install sudo reboot
In the menuconfig, I added support for Lucent WaveLan in Wireless,
Wireless LAN support, AT-T WaveLan Hermes Chipset and Hermes PCMCIA. There
was good setup info for Dells and Orinoco WiFi cards at this site. I also managed to find a
cached google web page that specified some port ranges that I needed to
exclude in /etc/pcmcia/config.opts. I changed the line...
include port 0x100-0x4ff
to include port 0x300-0x4ff
and added the line exclude port 0x100-0x2ff
This worked wonderfully! I see good messages about the card in the syslog
now. (Since originally did this, I found another
page that gives a good description of what needs to be done.)
I also edited /etc/wireless.opts and commented out the *,*,*,*)
lines so the "Wavelan IEEE example" section would pick up the wireless card.
In that section, I uncommented the RATE="auto" line and added
DHCP="y"
NICKNAME="gouda"
The machine worked at school (on an open WiFi connection). Yay!