LinuxQuestions.org - Blogs - linux-related notes by the dsc

Getting out of a dependency hell without reinstalling

the dsc — Thu, 18 Apr 2013 05:07:19 GMT

I'm not suggesting it's really preferable in any situation or in any specific situation, I just found myself in a "what the hell" point and thought it worth a try, and it worked.

I was just thinking that it would be nice that for slideshow videos there was a variable frame rate, and turns out that it exist. I've found that the program handbrake can handle that easily, so I decided to install. Unfortunatelly it's only on Debians unnofficial multimedia repository, which can have some serious conflicts with official packages. That's what happened after I naively upgraded just one or another additional package that I thought that wouldn't cause the same sorts of bugs I already had once, but it did. These are packages somewhat indirectly related with (g)mplayer, like libavcodec53 and related packages.

Once you install the multimedia repo versions, you can't mark to uninstall them on synaptic without marking to uninstall also something like 300 other packages, which would leave your system just with bare-bones x-server or something. Even though most of these packages haven't been upgraded, so they theoretically couldn't depend on those new packages "for real".

And they don't. With some guesswork it's possible to uninstall manually the newly upgraded packages and reinstall the official versions, all with "dpkg -r --ignore-depends " magic.

This is NOT a "tutorial", this is reference for myself just in case I do something stupid again in the future, or in case I stumble in some collateral problem. Try anything at your own risk, don't blame me. There's even lots of commands that just gave error outputs, as I haven't cleaned it up yet. Downloading of packages via web-browser usually precedes a "dpkg -i", which requires a previously downloaded package.

Code:

00:09:01 - root /home/d # dpkg -i libavcodec-extra-53_0.8.6-1_i386.deb 
Selecting previously unselected package libavcodec-extra-53:i386.
dpkg: regarding libavcodec-extra-53_0.8.6-1_i386.deb containing libavcodec-extra-53:i386:
 libavcodec-extra-53:i386 breaks libavcodec53
  libavcodec53:i386 (version 7:0.10.3-dmo1) is present and installed.

dpkg: error processing libavcodec-extra-53_0.8.6-1_i386.deb (--install):
 installing libavcodec-extra-53:i386 would break libavcodec53:i386, and
 deconfiguration is not permitted (--auto-deconfigure might help)
Errors were encountered while processing:
 libavcodec-extra-53_0.8.6-1_i386.deb
01:16:38 - root /home/d # dpkg -i libavcodec-extra-53_0.8.6-1_i386.deb --auto-deconfigure
dpkg: regarding libavcodec-extra-53_0.8.6-1_i386.deb containing libavcodec-extra-53:i386:
 libavcodec-extra-53:i386 breaks libavcodec53
  libavcodec53:i386 (version 7:0.10.3-dmo1) is present and installed.

dpkg: error processing libavcodec-extra-53_0.8.6-1_i386.deb (--install):
 installing libavcodec-extra-53:i386 would break libavcodec53:i386, and
 deconfiguration is not permitted (--auto-deconfigure might help)
dpkg: error processing --auto-deconfigure (--install):
 cannot access archive: No such file or directory
Errors were encountered while processing:
 libavcodec-extra-53_0.8.6-1_i386.deb
 --auto-deconfigure
01:18:38 - root /home/d # dpkg --auto-deconfigure -i libavcodec-extra-53_0.8.6-1_i386.deb
dpkg: considering deconfiguration of libavcodec53:i386, which would be broken by installation of libavcodec-extra-53:i386 ...
dpkg: yes, will deconfigure libavcodec53:i386 (broken by libavcodec-extra-53:i386)
dpkg: regarding libavcodec-extra-53_0.8.6-1_i386.deb containing libavcodec-extra-53:i386:
 libavcodec54:i386 conflicts with libavcodec-extra-53
  libavcodec-extra-53:i386 (version 6:0.8.6-1) is to be installed.

dpkg: error processing libavcodec-extra-53_0.8.6-1_i386.deb (--install):
 conflicting packages - not installing libavcodec-extra-53:i386
Errors were encountered while processing:
 libavcodec-extra-53_0.8.6-1_i386.deb
01:18:58 - root /home/d # dpkg -r libavcodex54:i386
dpkg: warning: ignoring request to remove libavcodex54 which isn't installed
01:19:38 - root /home/d # dpkg -r libavcodec54:i386
dpkg: dependency problems prevent removal of libavcodec54:i386:
 vlc depends on libavcodec54 (>= 8:1.0.4).
 libavformat54:i386 depends on libavcodec54 (>= 8:1.0.0).
 libavfilter2:i386 depends on libavcodec54 (>= 7:0.11.2).
 libavfilter3:i386 depends on libavcodec54 (>= 8:1.0.0).
 vlc-nox depends on libavcodec54 (>= 8:1.0.4).
 libmlt5:i386 depends on libavcodec54 (>= 8:1.0.2).
 libavdevice54:i386 depends on libavcodec54 (>= 8:1.0.0).
 ffmpeg depends on libavcodec54 (>= 8:1.0.0).

dpkg: error processing libavcodec54:i386 (--remove):
 dependency problems - not removing
Errors were encountered while processing:
 libavcodec54:i386
01:19:46 - root /home/d # dpkg -r --ignore-depends libavcodec54:i386
dpkg: error: --remove needs at least one package name argument

Type dpkg --help for help about installing and deinstalling packages[*];
Use `dselect' or `aptitude' for user-friendly package management;
Type dpkg -Dhelp for a list of dpkg debug flag values;
Type dpkg --force-help for a list of forcing options;
Type dpkg-deb --help for help about manipulating *.deb files;

Options marked[*] produce a lot of output - pipe it through `less' or `more' !
01:20:30 - root /home/d # dpkg --ignore-depends -r libavcodec54:i386
dpkg: error: --ignore-depends needs a valid package name but '-r' is not: illegal package name in specifier '-r': must start with an alphanumeric character
01:20:43 - root /home/d # dpkg --ignore-depends libavcodec54:i386 -r libavcodec54:i386
(Reading database ... 266993 files and directories currently installed.)
Removing libavcodec54:i386 ...
01:21:40 - root /home/d # dpkg -i libavcodec-extra-53_0.8.6-1_i386.deb 
dpkg: regarding libavcodec-extra-53_0.8.6-1_i386.deb containing libavcodec-extra-53:i386:
 libavcodec-extra-53:i386 breaks libavcodec53
  libavcodec53:i386 (version 7:0.10.3-dmo1) is present and installed.

dpkg: error processing libavcodec-extra-53_0.8.6-1_i386.deb (--install):
 installing libavcodec-extra-53:i386 would break libavcodec53:i386, and
 deconfiguration is not permitted (--auto-deconfigure might help)
Errors were encountered while processing:
 libavcodec-extra-53_0.8.6-1_i386.deb
01:21:52 - root /home/d # 
01:22:20 - root /home/d # 
01:22:20 - root /home/d # 
01:22:20 - root /home/d # dpkg --ignore-depends libavcodec53:i386 -r libavcodec53:i386
(Reading database ... 266985 files and directories currently installed.)
Removing libavcodec53:i386 ...
01:22:48 - root /home/d # 
01:22:48 - root /home/d # 
01:22:49 - root /home/d # dpkg -i libavcodec-extra-53_0.8.6-1_i386.deb (Reading database ... 266978 files and directories currently installed.)
Unpacking libavcodec-extra-53:i386 (from libavcodec-extra-53_0.8.6-1_i386.deb) ...
Setting up libavcodec-extra-53:i386 (6:0.8.6-1) ...
01:22:56 - root /home/d # apt-get download mplayer
Get:1 Downloading mplayer 2:1.0~rc4.dfsg1+svn34540-1+b2 [2,885 kB]
Fetched 2,885 kB in 24s (119 kB/s)                                             
01:29:00 - root /home/d # dpkg -i mplayer
dpkg: error processing mplayer (--install):
 cannot access archive: No such file or directory
Errors were encountered while processing:
 mplayer
01:29:07 - root /home/d # dpkg -i mplayer_1.0~rc4.dfsg1+svn34540-1+b2_i386.deb 
(Reading database ... 266988 files and directories currently installed.)
Preparing to replace mplayer 2:1.0~rc4.dfsg1+svn34540-1+b2 (using mplayer_1.0~rc4.dfsg1+svn34540-1+b2_i386.deb) ...
Unpacking replacement mplayer ...
Replaced by files in installed package mencoder ...
Setting up mplayer (2:1.0~rc4.dfsg1+svn34540-1+b2) ...
Processing triggers for mime-support ...
Processing triggers for man-db ...
01:29:41 - root /home/d # 
01:30:20 - root /home/d # 
01:30:20 - root /home/d # dpkg --reconfigure mplayer
dpkg: error: unknown option --reconfigure

Type dpkg --help for help about installing and deinstalling packages[*];
Use `dselect' or `aptitude' for user-friendly package management;
Type dpkg -Dhelp for a list of dpkg debug flag values;
Type dpkg --force-help for a list of forcing options;
Type dpkg-deb --help for help about manipulating *.deb files;

Options marked[*] produce a lot of output - pipe it through `less' or `more' !
01:30:35 - root /home/d # dpkg-reconfigure mplayer
01:30:48 - root /home/d # dpkg-reconfigure libavcodec-extra-53
01:31:02 - root /home/d # dpkg-reconfigure mplayer
01:31:05 - root /home/d # 
01:32:31 - root /home/d # 
01:32:31 - root /home/d # dpkg -i libavformat-extra-53_0.8.6-1_all.deb 
Selecting previously unselected package libavformat-extra-53.
dpkg: regarding libavformat-extra-53_0.8.6-1_all.deb containing libavformat-extra-53:
 libavformat54:i386 conflicts with libavformat-extra-53
  libavformat-extra-53 (version 6:0.8.6-1) is to be installed.

dpkg: error processing libavformat-extra-53_0.8.6-1_all.deb (--install):
 conflicting packages - not installing libavformat-extra-53
Errors were encountered while processing:
 libavformat-extra-53_0.8.6-1_all.deb
01:32:39 - root /home/d # dpkg --ignore-depends libavformat54:i386 -r libavformat54:i386
(Reading database ... 266987 files and directories currently installed.)
Removing libavformat54:i386 ...
01:34:13 - root /home/d # 
01:34:13 - root /home/d # 
01:34:13 - root /home/d # dpkg -i libavformat-extra-53_0.8.6-1_all.deb dpkg: regarding libavformat-extra-53_0.8.6-1_all.deb containing libavformat-extra-53:
 libavformat53:i386 conflicts with libavformat-extra-53
  libavformat-extra-53 (version 6:0.8.6-1) is to be installed.

dpkg: error processing libavformat-extra-53_0.8.6-1_all.deb (--install):
 conflicting packages - not installing libavformat-extra-53
Errors were encountered while processing:
 libavformat-extra-53_0.8.6-1_all.deb
01:34:20 - root /home/d # dpkg --ignore-depends libavformat53:i386 -r libavformat53:i386
(Reading database ... 266979 files and directories currently installed.)
Removing libavformat53:i386 ...
01:34:36 - root /home/d # dpkg -i libavformat-extra-53_0.8.6-1_all.deb (Reading database ... 266972 files and directories currently installed.)
Unpacking libavformat-extra-53 (from libavformat-extra-53_0.8.6-1_all.deb) ...
dpkg: dependency problems prevent configuration of libavformat-extra-53:
 libavformat-extra-53 depends on libavformat53; however:
  Package libavformat53:i386 is not installed.

dpkg: error processing libavformat-extra-53 (--install):
 dependency problems - leaving unconfigured
Errors were encountered while processing:
 libavformat-extra-53
01:34:39 - root /home/d # dpkg -i libavformat53_0.8.6-1_i386.deb 
Selecting previously unselected package libavformat53:i386.
dpkg: considering removing libavformat-extra-53 in favour of libavformat53:i386 ...
dpkg: libavformat-extra-53 is not properly installed; ignoring any dependencies on it
dpkg: yes, will remove libavformat-extra-53 in favour of libavformat53:i386
(Reading database ... 266977 files and directories currently installed.)
Unpacking libavformat53:i386 (from libavformat53_0.8.6-1_i386.deb) ...
Setting up libavformat53:i386 (6:0.8.6-1) ...
01:36:05 - root /home/d # dpkg -i libavformat-extra-53_0.8.6-1_all.deb 
Selecting previously unselected package libavformat-extra-53.
(Reading database ... 266983 files and directories currently installed.)
Unpacking libavformat-extra-53 (from libavformat-extra-53_0.8.6-1_all.deb) ...
Setting up libavformat-extra-53 (6:0.8.6-1) ...
01:36:11 - root /home/d # dpkg-reconfigure mplayer

Mplayer and VLC were not playing .flv files before that, but I managed to make it work by repeating the same sort of process with one more package (libavutils50 or something). It doesn't do that alone though, I can't simply reinstall the official package normally, as there are still quite a few broken packages, something like 20. But these are more obviously only video/multimedia-related, and then I could mark to uninstall them all on synaptic, uninstall, then install only those again (mostly by virtue of simply being a dependency of something like VLC, no need to add them individually for the most part). For some reason sometimes I have to click on "apply" once, I'll receive some error message related with broken packages, but it will work when I click on apply for a second time, without changing anything.

A good $PS1 for displaying long paths in the bash prompt

the dsc — Thu, 14 Mar 2013 04:29:10 GMT

Code:

PS1='${debian_chroot:+($debian_chroot)}\t \u $(if [ ${#PWD} -gt 30 ] ; then myPWD="${PWD:0:12}�/\W" ; myPWD="${myPWD:0:25}�" ; else myPWD=$PWD; fi ; echo $myPWD) $ '

It will display full paths smaller than 30 characters entirely, but for longer paths it will get only the first characters of the whole path, and the fist characters of the last folder in the full path.

The end result is something like:

HH:MM:SS tuxguy /home/tuxguy/mystuf.../xmas pics $
HH:MM:SS tuxguy /home/tuxguy/mystuf.../long folder nam... $

It's a combination of what I already had (everything up to "\u") and something I saw here, in the comment section:

http://www.cyberciti.biz/tips/howto-...up-prompt.html

It was using a somewhat different method that didn't work for me, perhaps because of different/wrong quote characters, environment variables, or whatever.. So I stuck it all into PS1, ditching the "PROMPT_COMMAND" part that I didn't get. And it worked.

I like having the first characters of the current folder better, they're often more informative than the last ones I guess.

Ditch Deja Vu Sans Condensed, use Liberation Sans, and be happy

the dsc — Sun, 10 Mar 2013 00:01:02 GMT

I used to like the condensed variant of Deja Vu, which is the default font (in the regular variation) for many linux distributions/DEs I guess. I've always disliked the regular variant, I find it too wide, while the condensed is very nice to look at. It's quite a hassle to actually use it everywhere though, either the font is buggy, or the apps that read it are buggy. I think it's the former as other fonts with two "nested subtypes/variants" work (such as liberation). The GUI settings editors won't be able to set it, picking either "book" or "semi-condensed" variants instead, if at all.

In some cases (KDE, and perhaps GTK3 or GTK2 (not both)), you can edit the config files manually and have it working. Which is OK if you only use mostly KDE, but if you use a little bit of everything and wants to make it all look reasonably homogeneous, both in theming, fonts, and font antialiasing/hinting, it may be quite troublesome, as the GUI settings editors will tend to change things you didn't want to change, so you go on one by one fixing everything, but at each time it's breaking some other config you had previously set, sometimes not immediately obvious as running instances will remain running with the previous/correct settings sometimes, and sometimes (depending on the settings editor and/or widget set) it will immediately "reset" the theme/font settings even on running instances.

Liberation sans is a very acceptable replacement. It's already narrow enough in its regular variant, the narrow variant is a bit too narrow. Every GUI settings editor will understand it, every widget set will accept it, and so then there's one thing less to manually fix on the config files.

Make some apps (mainly KDE apps with text fields) start faster

the dsc — Wed, 27 Feb 2013 01:37:18 GMT

Just create a folder "~/.compose-cache".

Or better yet, at least in a semi-philosophical level, have your startup script to create "/dev/shm/.compose-cache" every time, and just create it once, and a soft link to your home folder. No unnecessary access to your disk, writing to and reading from memory instead. Which is often assumed to be better for several cache-type usages, both for performance and (supposedly) for the lifespan of your hard disk*. Some people even make some scripts that will copy the entire profile folder of an application such as firefox or chrome to /dev/shm, where the browser will use it, and synchronize it back to the disk when it is closed.

If the PC is shared by more people who may use it with simultaneous sessions, perhaps it's better to have it as "/dev/shm/$USER/.compose-cache" instead.

* however, some studies surprisingly show no significant correlation between heavy disk usage and decreased lifespan, perhaps even somewhat of an inverse correlation when lower usage is accompanied by lower temperatures, which perhaps even more counter intuitively, correlate more with shorter lifespan than hotter temperatures. I'm not suggesting you should light a fire inside your cabinet though.

Controlling VLC via openbox key bindings... and wget(!)

the dsc — Mon, 28 Jan 2013 16:21:38 GMT

On openbox' rc.xml, in the corresponding section:

Code:

    
      
wget http://127.0.0.1:8080/requests/status.xml?command=pl_previous -O /dev/null 
      
    
    
      
	wget http://127.0.0.1:8080/requests/status.xml?command=pl_next -O /dev/null

With "web" interface enabled

By far the weirdest workaround I've seen on this sort of thing. If that's really a workaround and not the "proper" way to do it.

Command line has options such as:

vlc --global-key-prev p

which sort of works, except that in this case it won't be really the previous, but just a random element. I guess the playlist is being re-loaded, but it won't work if you don't give such mysterious "string", which is supposed to be a file, according to the error console. VLC's instrunctions are as follow:

Quote:

--global-key-prev= Previous
Select the hotkey to use to skip to the previous item in the playlist.
--key-prev= Previous
Select the hotkey to use to skip to the previous item in the playlist.

Which to be honest I can't really tell if that's a command line to navigate through the playlist, or just to set the hotkeys. It sounds more like the latter, but that's a rather odd command to have on command line begin with. Furthermore I didn't succeed in trying to re-set the hotkeys that way either, so I guess it's really a navigation control, as it's something more expected to have on command line, I guess.

Wacom tablet and the "keepshape" option, on openbox

the dsc — Thu, 22 Nov 2012 03:54:42 GMT

On KDE you can set "keepshape" on "systemsettings", but apparently there's no way to do it in any other DE. The proper way to have "keepshape" working does not seem to work on Debian Wheezy (setting it on xorg.conf.d/50-wacom.conf). The "standard" alternatives are a static xorg.conf, and perhaps figuring hal/udev rules, which I don't know how to do.

Another way to do it is via xsetwacom, explicitly setting the active area along the lines of:

Code:

xsetwacom set "Wacom Bamboo stylus" Area 0 0 11531 9225
xsetwacom set "Wacom Bamboo cursor" Area 0 0 11531 9225
xsetwacom set "Wacom Bamboo eraser" Area 0 0 11531 9225

These lines are to be placed in an autostart script appropriate for the DE you're using.

Instead of "Wacom Bamboo whatever" yours might have some other name. In order to get the proper name, you should use the command "xsetwacom --list devices", as a normal user (or at least that worked for me).

I got these numbers, the coordinates ("0 0 11531 9225") from KDE's wacom settings panel. Over there there is a section where you can set the tablet to be proportional to the display, and move the active area as you wish. "0,0" means that in my case it starts exactly at the top right. You get these numbers from the "diagram"/what-you-see-is-what-you-get thing.

Setting it on KDE will work only for KDE as it require a KDE service to be running. If you have some KDE installed you don't need to log on a session to get these values, just starting the service on systemsettings (startup and shutdown → service manager → select "wacom tablet" and click on "start") will suffice. Perhaps one could even manage to have the service working in some other DE via some of those dcop/dbus commands I guess, so the settings panel would be totally functional in any DE you can make the service run at startup.

Perhaps there's a less newbie-ish way of getting these coordinates, but I don't know how to do it, and it just works this way, so that's enough for me. I saw somewhere even a script that would generate it every time I guess, but I didn't looked carefully at it yet:

Quote:

Originally Posted by maxy

As far as I know, at the moment, you have to configure this manually. I have the opposite problem: my screen is wide and the tablet "normal". I also prefer to have a bit a smaller active area. I use the bash script below to map the bottom area of my tablet. Obviously you would have to change it for your needs.

Code:

stylus="Wacom Intuos3 6x8"

(( w = 40640 )) # intuos active area
(( h = 30480 )) # intuos active area
# 1920 1080 <-- wide screen (16:9)
(( w2 = w*6/10 )) # use only 60% of the tablet width (free choice)
(( h2 = w2*9/16 )) # calculate the height to make the ratio correct
(( topx = 0 )) # active area starts at the bottom left of the tablet
(( topy = h - h2 ))
(( bottomx = topx + w2 ))
(( bottomy = topy + h2 ))

xsetwacom set "$stylus" topx $topx
xsetwacom set "$stylus" topy $topy
xsetwacom set "$stylus" bottomx $bottomx
xsetwacom set "$stylus" bottomy $bottomy

Here's the link:
http://forum.intilinux.com/mypaint-h...g7663/#msg7663

If the xsetwacom's syntax given above is outdated, I think you just need to use the same variables on a line like "xsetwacom set '$stylus' Area $topx,$topy $bottomx, $bottomy", or perhaps vice-versa.

Gmplayer/mplayer-gui standard package works perfectly on Debian testing (Wheezy)

the dsc — Wed, 21 Nov 2012 18:48:07 GMT

If you like mplayer's native GUI, and you're about to install Debian testing (currently Wheezy), avoid using packages from deb-multimedia as much as possible.

To me it seems that the problem is actually with ffmpeg (deb-multimedia's), but I'm not willing to test it. I think it somehow interferes with mplayer-gui/gmplayer.

Just a few days ago I had an Wheezy install where I had gmplayer working, but very inconsistently, I would double click some video file and in five times out of fifteen it wouldn't work, showing dozens of error message windows instead. And it wasn't related with a particular type of file/container/codec, as it would work on the same file afterwards, sometimes immediately after. It happened despite of the fact that I was keeping outdated packages (the newest ones wouldn't work at all, I don't recall if mplayer and mplayer-gui were incompatible or what), after trying several ones until find the latest one that would work.

To my surprise as I've reinstalled Wheezy, mplayer and gmplayer were both working with a default install. But it didn't last long as I've also added deb-multimedia's repository and updated everything but mplayer and gmplayer. The worst part is that even though I suspected that ffmpeg was the offending package, there was some dependency-hell sort of situation that made "impossible" to downgrade it. So I've reinstalled Wheezy twice in two days. Yeah, I'm turning back my badge and my pendrives with live linux distros, both the one I carry on the belt and the one I carry on the ankle.

Apparently you can install avidemux (from deb-multimedia) and still have gmplayer working, or at least that's the situation before rebooting.

Bash quickie: loop N times, without doing math

the dsc — Tue, 23 Oct 2012 02:07:47 GMT

I thought that the only way to do such type of loop in Bash would require something like "n=$(($n+1))" within the loop, with the loop being conditioned to "n" adding up to whatever you want, such as "until ((n==30)) ; do". But you can actually do "for i in {1..20} ; do ; done". That's somewhat similar to Basic's "for i=1 to 20 ; whatever ; next i".

But it won't work with variables for the starting and ending numbers. If you want to use variables, you must use $(seq): "for n in $(seq $startnum $endnum) ; do ; done.

Pure-text de-truncator script, a work in progress

the dsc — Sun, 30 Sep 2012 18:28:22 GMT

Sometimes I want to quote some PDF on pure text, luckily enough it would have actual text rather than being an non-OCRed image, but often there's still the problem that the text on the PDF is truncated/formatted in a fake/dumb way, with actual "new lines" to break the page, which may not make the text completely unreadable when pasted in a text editor, but it's reasonably annoying, and may require quite some time to fix manually.

I'm trying to create a script or one-liner with a few pipes that would try to fix those texts. I think it basically has to read the text line by line, check if it's an empty line (in which case it does nothing I guess), if it ends in a period (again, do nothing), or if it's too short (and does nothing, just in case it's a header). If none of these apply, it's likely that it ends in a unnecessary "new line", which is then stripped. Not necessarily in this particular sequence of tests, perhaps there's some other sequence that's more logical for the flow.

So far what I have is:

Code:

xclip -o | while read a ; do lchr="${a#${a%?}})" ; if [ ! $lchr = "." ] ; then echo $a NODOT | tr -d '\n' ; else echo $a PERIOD ; fi ; done

Very preliminary, the "xclip" means that the text is coming from the clipboard, eventually I'd change it for some "cat $file/$*" phrasing I guess. The all-caps echos are obviously just a "debug mode", in order to make utterly clear what's going on.

For while it only checks if the line ends in period or not, and strips the "new line" if it does not. I still got to find out how to test empty lines and some arbitrary but reasonably good number of characters for eventual headers.

Any help is appreciated!

References:
http://www.unix.com/shell-programmin...er-string.html
http://www.linuxquestions.org/questi...ng-sed-191121/

Rant: what was wrong/"bad" with mplayer-gui/gmplayer?

the dsc — Mon, 03 Sep 2012 15:41:32 GMT

I don't get it. It wasn't like it inspired someone to say that it's the GUI that one ever dreamed about, but was fully functional, and funnily enough more reliable than all the alternative GUIs that used mplayer as a back-end, as far as I've looked at. I guess the decision eventually resulted in a (current to me at least) problem with subtitles/fontconfig, as subtitles will work on mplayer and mplayer2, but not on gmplayer, even though it never used to be that way before.

It's not with mplayer that this sort of thing happens though, I find very frustrating things like the "instant filter" on konqueror being removed with the reasoning that it's a "plugin" and it needed to be built-in, but they've done that before the built-in version was implemented, leaving the software less functional. I'm not a programmer but I think that it's not a requirement to remove the availability of such plugin in order to develop/adapt (from dolphin or from the "plugin") the built-in version. Why cripple the software instead of making this end-user-invisible/technical improvement without crippling it? If it isn't broken, don't remove it.

That remembers me somewhat of the whole polemic against Gnome 3. I'm happy I never used Gnome*, by the way. I just hope openbox remains forever more or less like it is today.

[/rant]

* from a non-former-user perspective, it does not look that bad though. I still wonder if things like gnome 2 and kde 3 should really have been "dead ends", rather their successors being independent "forks", and their predecessors being eventually phased out gradually as people moved to them, if they did. [rant again] Perhaps that's why. Perhaps developers want to see their new creations being widely adopted, and there's no better way than discontinuing or crippling alternatives/predecessors, with which perhaps they don't have the same sort of personal attachment that they may have with the new things they've created from scratch. [/rant]

Three convenient customizations to firefox / iceweasel

the dsc — Sat, 14 Jul 2012 21:40:21 GMT

Restore an old/previous session, but do not load the tabs right away: an end to the dilemma of empty session versus restoring tabs. Add to "about:config", if it's not already there:

Code:

browser.sessionstore.restore_on_demand

Set to "true", and it doesn't hurt to add also:

Code:

browser.sessionstore.max_concurrent_tabs

Set to 0. I guess it depends on the version or whatever, I don't care to find out. I wish I knew that long ago, apparently it exists ever since firefox 4. Which isn't that long ago anyway, with all the new-version-number-frenzy they got into.

Making mozilla look native: I guess it may depend on which DE you're using; I'm on openbox and using qtcurve, which generates the same version of a custom theme for both QT/KDE and GTK (but not GTK3), and it worked quite well, it only lacks a highlight on tabs as we hover:

http://userstyles.org/styles/49622/n...fox-4-on-linux

It says "4" but works for firefox 10 or whatever it's now.

And for those who prefer compact interfaces: "replacing the Iceweasel/firefox button with an icon":

http://glandium.org/blog/?p=1554

Mplayer configuration tips

the dsc — Sat, 07 Jul 2012 06:50:47 GMT

Setting the default volume for mplayer and gmplayer

Despite of having different config files, "config" and "gui.conf", this is one configuration that gmplayer inherits from "config", and will be ignored if set on "gui.conf", with the volume being reset to max and the line in the config file removed. That's one of those easter-eggs to give type As heart attacks. I think that's valid for Debian squeeze's gmplayer version, it may have changed afterward.

Code:

softvol = "yes"
af=volume=-24

"Softvol" stands for "software volume", meaning that mplayer will have its own "independent" volume relative to the OS global volume settings such as PCM, Master, Front, etc. I don't know why exactly the volume setting is negative, perhaps it's relative to some absolute default that is not zero, that just happened to be the value that sounded OK for me.

Controls for brightness, contrast and saturation
(The keyboard shortcuts are the numbers in the number row)

Just use "vo=gl" instead of "vo=xv" (default I guess), no noticeable compromise in performance. I don't know the requirements for that, but right now I just have standard wheezy and/or squeezy packages from Debian's repositories, not debian-multimedia.

Changing subtitles colors

Code:

ass=yes
#ass-color=FFFF00
ass-border-color=000000
subfont-text-scale = "7"

I don't get exactly what's the deal with selecting the color, because if I recall when I tried to set yellow the subtitles would be cyan, and so on, I was probably using the wrong color "codes", and I couldn't find reference for the proper one to use. The subtitles are somehow yellow with black contours with the configuration above, anyway. Gui.conf has no options on subtitles colors, and it will be yellow as well, perhaps that's the default for "Advanced SubStation Alpha".

Fast or slower tempo, preserving pitch -- plus normalization

Code:

af=scaletempo,volnorm=2:0.75,volume=2

The "2" on volnorm means that it will base its normalization on two samples or something like that, the default is one.

Default and alternative configs for gmplayer

If you read an alternative gmplayer config file, it will overwrite the default one. The best way to circumvent that I've thought so far is simply to set whatever defaults and mark the gui.conf file as write-protected. So you can call alternative configs whenever you want (via "-gui-include"), and the default settings will be preserved. You just got to remember you've done that once you decide to change some default configuration, so you un-protect it for a while.

Improving/fixing font rendering on Debian's web browsers

the dsc — Fri, 08 Jun 2012 20:59:15 GMT

Warning: I don't know the whole story about this (no warranties, you're on your own if you decided to follow these instructions), but after some recent upgrades my Debian install (testing/unstable) had awful font rendering on web browsers (everything nice everywhere else somehow). I like to "force" freesans (which is very somooth and anti-aliased) on almost everything on the web, and it was looking like some pixelated Times New Roman or Courier New.

I went googling for solutions and stumbled on some stuff. First I tried upgrading/downgrading some packages, apparently* with no effect whatsoever. What did seem to solve it was to download some config files and use those instead of the ones I had:

Quote:

[...]

Code:

 # wget http://www.osresources.com/files/centos-windows-fonts/fontconfig.tbz

Backup old configuration files then extract new ones into /etc/fonts.

Code:

# for i in `ls /etc/fonts/*.conf`; do cp $i $i.bak; done
# tar xvjpf fontconfig.tbz -C /etc/fonts/

Clean up.

Code:

# rm fontconfig.tbz

[...]

Then log out of the session and log in again.

I saw it in this blog post:
http://briolidz.wordpress.com/2012/0...nts-rendering/

Now I have freesans nicely rendered everywhere on the web again, to the disgrace of the web-designers who want to impose different fonts on me.

* Actually it may have worked, but I wouldn't know as I tried to restart just the web browser rather than the entire user session. So perhaps messing with packages was what really solved it for me, or was part of the solution. At very least it didn't seem to have created more trouble so far. If I recall I downgraded libcairo2, and upgraded libfontconfig, fontconfig and fontconfig-config. But after that I did what I saw on this blog, which may have undone one of these steps (like replacing fontconfig-config, I guess).

Pseudo-template noob hack that may apply for several situations

the dsc — Thu, 12 Apr 2012 21:10:33 GMT

There's a painting software, "mypaint", which is very nice, but it does not support templates, at least not yet.

I've created a template (just a "blank" file with the correct resolution and whatnot) on GIMP, saved it as "ora", but instead of opening this file manually every time, and remembering to save it with a new name in order to not overwrite the template, I came up with a script that will create a new name automatically, with no risk of overwriting the template.

Code:

template=$1

export DISPLAY=:0 # I'm not really sure it's needed or why it's there anymore

filename="$(date '+%y%b%d-%H%M').ext" # creates a filename based on the time and day you run the script

cp "/path/to/your/$template" "/dev/shm/$filename" && # just in case you don't end up saving the new work from the template, it copies it first to the memory

whatever-program "/dev/shm/$filename" &

while true ; do inotifywait /dev/shm/$filename && cp /dev/shm/$filename /path/to/destination/folder/ # if the file on /dev/sh (memory) is changed (when you save it), it's copied to a given destination folder on the hard drive.
  done

Then you can add some application menu entry pointing to this script, or even a submenu with several template options, given as parameters to the script ($1).

Got banned from the ubuntu forums

the dsc — Thu, 12 Apr 2012 20:24:41 GMT

First time I got banned from any forum, that's a historical moment! :)

I started a thread questioning whether the common forum practice of closing old threads when they're bumped, soliciting the users to create new ones, is really a good idea. I think it's not, that by not doing it the threads will be richer, containing more information, instead of being scattered around many topics all over the years.

Anyway, I referred to moderators who do that (not anyone in specific or even ubuntu's forum staff in particular) as "anti-bump nazis" with a strikeout line, followed by "moderators"... I guess this (rather than merely questioning the practice) was considered an insult to the staff, and for this reason the thread was blocked from further discussion, after two or three replies.

Wanting clear things out, I opened another thread, saying, "no insults intended", apologizing and expressing my surprise to the notion that such use of the term "anti-bump nazi" would really be offensive, and asking whether I could re-post the same point about bumping-or-not, with "anti-bump nazi" removed. The answer apparently is "no, you can't", since this second topic got deleted and I got banned forever, with no prior advertence or further communication...

My recommendation to anyone who may be reading it and posts over there every once in a while... write as formally as you can, with no jokes or anything, someone might get offended even if you never imagined it was even possible. :rolleyes:

Windows-like folder organization for binaries and libraries through symbolic links?

the dsc — Fri, 25 Nov 2011 21:28:35 GMT

Just another of those things I catch myself wondering sometimes, but I don't think that are important enough for the forums.

What if linux had folders for each program, more or less like on windows, but still had its "grab-bag" folders for libraries and whatever that may be used for more than one program, but instead of using it through a "$path", it had a symlink within the program's own folder?

Or perhaps it could use by default through something like "$path", but just in case you want to tweak the program a little bit you could place different versions of whatever in that folder (those would be prioritary), instead of having something like /opt, which seems to me a bit more complicated to do and tweak.

But I don't know really, perhaps that's just dumb, useless and gross. It would at least require a major re-planning of the whole folder structure standard, which is a good reason not to do it already I guess, it's not like not having it is hindering something very important. There's a linux distro that is attempting to do something more or less like it I guess, I don't know if it's "exactly" like I've described though.

High CPU usage + underclock vs. low CPU + normal/high clock

the dsc — Sun, 13 Nov 2011 20:02:53 GMT

This is just a question rather than a typical blog post. I know blogs aren't really supposed to be used for that, but I think it's justifiable since it's not that important or urgent, even though it has some practical applications. So I'm not "stealing" attention from other users with more pressing matters over there.

The question is basically "what gets more work done":

a) higher CPU usage per process (limiting less or not limiting the usage at all with tools such as cpulimit) but in the other hand "underclocking" a little bit (with things like cpufreqd or cpufreq-set) so that the CPU does not melt

b) having the CPU frequency next to its maximum or a bit above average, but limiting more the load of some specific applications with cpulimit.

I think that there may be advantages and disadvantages for each strategy, so perhaps it depends on what is that one is using the PC at the moment. For example, I imagine that the "b" strategy leaves the user and other processes with extra responsivity/resources, whereas "a" privileges the high-load CPU process somewhat, and everything else may be a bit sluggish, even though it's still useful for minimizing CPU temperature and maybe power consumption (even though that part is a bit complicated and sometimes a higher performance may result in less consumption in the long run, or something like that - see here and here some indirect statements with links to the original source (apparently offline momentarily) by the Intel's kernel developer, Arjan van de Ven, saying that the "ondemand" governor may be more efficient in terms of power-saving than "conservative" and even "powersave" governors)

But perhaps "a" may be more inefficient even when for some reason you want to privilege one process in detriment of the others (such as running something like folding@home when you're not using the PC), perhaps this process could get more work done even if it were more limited by cpulimit, as long as it could work in a less underclocked environment. But of course could be the other way around, perhaps this extra Mhz don't mean much, what matters most is being less restricted even if the CPU is operating on lower frequencies.

And, of course, could be that it does not make any difference any way or the other, or that it depends more on the fine tuning of "a" and "b" strategies than the strategies per se.

So, if anyone has a more enlightened perspective on the subject, I'd be curious to hear it.

Commanding openbox through scripts or command line

the dsc — Thu, 20 Oct 2011 03:03:28 GMT

Is it possible? If it is, it's not obvious in any way, nor documented in the manual.

I think that would be an interesting feature if it currently lacks it. I don't really know, but I think it wouldn't be so much of a huge thing to implement the possibility of running commands such as "openbox-command menu-id-whatever" to make a given menu pop-up; or "openbox-command nexdesktop" so a script could move the view to a different desktop (but perhaps it can be done with wmctrl or something else). Then openbox could be used also as something like gxmessage/zenity/kdialog, or interact with things such as "brightside", among other possibilities that didn't occur to me yet.

Originally thought as a question on the actual forum, but I'm pretty sure the answer is no, so I'll just post it here anyway. Someone may read and think it's a good idea and make a feature request*. Or just comment saying it is possible already, and explaining how to do it.

* - if you would do it, request "window grouping/tabbing" too. Their rationale for not having it does not make sense -- because it's not just a matter of compensating for bad GUI design in one application, since as a DM/WM feature, it is not restricted to grouping windows of a single application in a single window. In fact, in a way it's not that different from "cascading" windows in a more convenient way. Also, by having that, it's halfway into eliminating the "need" for taskbars like tint2 or fb/spanel, as the title bar can be in effect turned into a taskbar. Perhaps even automatically if it wouldn't be to messy to have an option like "group with next window", which could be set concurrently with "lower" for the minimize button. The only difference is that it would be so almost identical only when the window is maximized. When it's not, it's somewhat as if you had tiny virtual desktops, each one with its own taskbar.

And "keychains" for mousebinds would be neat too.

A very tiny quick dictionary look up script

the dsc — Thu, 13 Oct 2011 03:05:47 GMT

For a long time I had fantasdic set to respond to a desktop environment keybinding, it would then appear with the result for a query that was made through the clipboard. I don't recall how much of it was its own functionality and how much was some sort of wrap-around scriptery. I'm so used to use computer dictionaries in that way (instead of directly going to them and typing) that to me it's almost useless without this keybinding functionality. And indeed it has been quite some time since it does not work quite well for some reason.

It works only for the first time the fantasdic is called, then in the second one (when it's already on the system tray), its window wouldn't show up, and there would be some X-related error in xsession-errors. I don't remember if I've ever made a bug report of some sort, I remember having mentioned it somewhere though. As a matter of fact, I'm not even sure it's still the case.

Nevertheless, I've made this tiny script to make up a little bit for the virtual "uselessness" of fantasdic:

Code:

#! /bin/bash
query=$(xclip -o)
dict -h localhost "$query" | sed 's#^\(.*From.*\)# \1 #' >/dev/shm/dict.tmp
kdialog --title "$query" --textbox /dev/shm/dict.tmp 400 550

With an openbox keybinding, it will pop up a kdialog text box with even a little text formatting (just bold "header" lines for each dictionary). You need only to select the text, doesn't need to "copy" it, just do whatever is the keybinding you set to run the script. It hasn't all the functionality that fantasdic had (like double clicking any word on the text/definition box and then going to the definition of that word), but I think it will be a decent replacement.

Besides kdialog and xclip (which gives the clipboard as input), I also have dict and dictd (the former is a client and the latter a server, if I'm not mistaken; I don't remember how much they're independent from each other), and dictionaries installed locally (hence "localhost"), but it could work with remote dictionaries as well, as long as someone fixes the dict query accordingly.

"Wait until a process ends, a download ends, or a file conversion ends, then do X"

the dsc — Tue, 30 Aug 2011 20:06:41 GMT

Sometimes you want the computer to shut down after you've finished downloading something or converting some large file into some other format, but you want to go out or go to sleep instead of waiting and checking for it yourself.

For these situations I've come up with two tiny scripts. The first one waits until a file hasn't changed its size on three consecutive intervals of twenty seconds:

Code:

a=0 ; file="$1" ; if [ -f "$file.part" ] ; then file="$file.part" ; fi ; until (
(a==3)) ; do if [ ! -f "$file" ] ; then exit 0 ; fi ; size1=$(stat -c%s "$file")
 ; echo $size1 ; sleep 20s ; size2=$(stat -c%s "$file") ; echo $size2 ; if [ $si
ze1 == $size2 ] ; then a=$((a+1)) ; echo $a ; else a=0 ; fi ; done
exit 0

You'd use it like:

waitfile &&

(without the less/more-than brackets)

The point of the whole checking on the existence of a "$file.part" is to have it compatible with firefox' downloads without having to bother remembering to explicitly point it to the temporary file that firefox uses. If such file exists, it assumes it's dealing with a firefox' download, and thus checks the .part file, not the file with the actual name, which would have a "zero" size for a good while. But it's not specific to firefox' downloads at all.

The next one waits until a process no longer exists:

Code:

until [[ ! `ps -A | grep $1` ]] ; do echo "$1 still running" ; sleep 5 ; done

The "syntax" is the same:

waitfor &&

I guess that this one can be a bit tricky sometimes depending on what you have running. Perhaps there will be more than one result for the same "grep" on "ps -A", so it could theoretically wind up waiting for more than one process, not only the one you have in mind. And this other process could "never" finish for whatever reason. An example that comes to my mind is a paused instance of mplayer (which wouldn't ever exit) or gmplayer (which does not exit as it reaches the end of the file), running concurrently with an instance of mplayer (which exits when it reaches the end of whatever it was playing, when it's not on slave mode).

Fixing the annoying firefox' associations issue once and for all

the dsc — Fri, 19 Aug 2011 20:49:38 GMT

Probably affects all gecko-derived browsers, like iceape, kazehakase, epiphany, namoroka, whatever. May be exclusively a KDE-derived problem though, I don't know.

Sometimes, for some reason, the system decides that you probably want to open downloaded PDFs, from iceweasel's/firefox' downloads window, with krita or gimp, programs that won't open PDFs at all. It's really confusing because you even have a correct, pdf-reader application set as default in other parts of the system (file managers, both KDE's and Gnome's), and you even get it from "xdg-mime query default application/pdf". Then you conclude that it must be firefox' sole fault, and you go and find some preferences related with handling files. But these will affect only how it will handle files "on the fly", files it's about to download/already cacheing, not which program will be used to open the file on the downloads window.

As it happens, linux has some apparently redundant file-association settings, firefox seems to pick a "unusual" one, whose very reason to exist may be questionable.

Here's the solution, though:

Quote:

Originally Posted by the black cat

Re: Gimp default for opening PDF's from Firefox

I got a solution that works from another forum (see here). I have copied it below:

to make pdf open using a specific application when you double click it in the FF download manager, edit /usr/share/applications/mimeinfo.cache , search for: application/pdf= . Put the name of *application*.desktop of the desired application right after = .

So to use okular, it should look like this:
application/pdf=kde4-okularApplication_pdf.desktop

Notice the extra kde4- , this is necessary because okularApplication_pdf.desktop is in the subfolder kde4 in /usr/share/applications.

Replace okularApplication_pdf.desktop with whatever name of the .desktop file of the app you want. If it's in the same folder as mimeinfo.cache, i.e. /usr/share/applications then use the name of the .desktop folder, if it's in a subfolder then prefix it with kde4- , or kde- , whatever. See if this works.

mimeinfo.cache is a dynamic file and will be changed automatically when you install/uninstall a programme (this explains what happened when you uninstalled gimp). Of course you can keep a local copy of this file maintaining your own associations across reinstalls), you need to put this copy in ~/.local/share/applications , also a copy of /usr/share/applications/defaults.list

Source: http://forums.opensuse.org/english/g...ml#post1945902

I wonder how these people find out how to fix this sort of thing among all the mess that linux configuration files are.

As a sidenote, I didn't have to copy this "defaults.list" file, which does not even exist on my system (debian wheezy/sid, testing). Otherwise it works, and you don't even need to restart firefox.

Emerge + yaourt + apt-build/checkinstall + debtorrent/apt-p2p

the dsc — Sat, 16 Apr 2011 21:44:13 GMT

Just an idea I had sometime ago, perhaps some bored developer sees it and might want to make it a reality.

It would be quite interesting if such a compiling/packaging system could be made. You compile with whatever flags you want and whatnot just like Gentoo's emerge, then it is converted into a package like apt-build or checkinstall (if emerge doesn't do that already), and then this package is seeded in a community-built meta-repository with many alternatives (somewhat like Arch's yaourt), but instead of (maybe concomitantly) being hosted in a conventional repository somewhere, it's seeded via some p2p protocol (like debtorrent and/or apt-p2p).

If there were enough people using it, they would have the benefits of both packages and compiled software, at least inasmuch they share similar hardware and compilation preferences. Even if you don't find exactly the same configuration parameters that you'd like, this chimeric system could offer you a pool of available options that are compatible with your hardware at least (somewhat akin to yaourt here). If none pleases you, then you got to compile according to your picky tastes. And you start to share it automatically, perhaps influencing more people on your choices.

Even those with more "weird" hardware would start to seed their packages, so as long as there's not only a single person with a given hardware, they could still share compiled packages even though they're more scant, as there are fewer people building them.

It could be even further hybridized with the conventional repository systems. When there are no available options for a given package, as you finish compiling something, your program talks with the server counterpart somewhere, which then copies it from you and starts to seed it along with you. So everybody has an increased chance of actually downloading the compiled package, as it does not depend of some seeder being online at the moment.

And there could be somewhat of a twist here: perhaps there could be a given disk space that people could offer to host packages that they don't even have chosen to install themselves, just for the sake of increasing the availability somewhat as if there was a central server, but instead it's "just" some swarm intelligence type of thing. The software decides what to host and what to delete based on the availability in the network.

Single, "teleporting" instance of system monitor (or something else)

the dsc — Wed, 30 Mar 2011 06:20:45 GMT

Code:

[ -z $(ps -A | grep ksysguard) ] && ksysguard & disown &&
wmctrl -iR $(wmctrl -l | grep "System Monitor" | sed 's/\ .*//')

Depend on wmctrl.

Are you tired of winding up with multiple instances of a program of which you don't really need multiple instances? Your problems are over! Now you can just press the hot-keys or whatever shortcuts you have, without ever worrying about earlier instances. No more sore joints for having to close THOUSANDS of instances that hog all your RAM! No more cluttered desktops filled with multiple instances of programs of which you don't really need so many instances!

With a script like this set as an alias or just being the target of a hotkey/shortcut, these are all things of the past!

You got to use wmctrl -l in order to check what could be a constant on the window name of a program (in this case, "System Monitor"), as it can only rarely rely on PIDs, using some obscure internal reference instead, which is "grepped" with grep and sed.

WARNING: may cause some wmctrl error messages on .xsession-errors each time it runs the program "for the first time" (as opposed to bringing the extant instance to the current desktop).

Solved: "mount error: could not resolve address for . No address associated with hostname"

the dsc — Thu, 06 Jan 2011 00:19:50 GMT

This error may happen when trying to mount a cifs/smbfs share, be it either via fstab or command line. Most people seem to either not have the problem or circumvent it by using static IPs instead of hostnames, which will work only as long as the IPs don't change. Funnily enough, it seems that for some people the problem is exactly the other way around, IPs won't work, only the hostname. Weird. This is also kind of funny/infuriating as the share will be readily accessible via konqueror, and perhaps some other file managers that can reach samba shares, and the name will be there.

Anyway, the way I found to fix it for me was to add "wins" on the /etc/nsswitch.conf file. I had "wins" previously installed for some reason, probably due to some earlier attempt at doing the same thing, but as it seems, the installation process does not change the configuration files as needed to actually work.

Here's the line I added "wins" to:

Code:

hosts:          files mdns4_minimal [NOTFOUND=return] dns wins mdns4

And here's where I found it: http://ubuntu-virginia.ubuntuforums....36&postcount=8 (in Spanish)

Nearly at the end of this lengthy post.

Apparently "update-rc.d" does something essential

the dsc — Sat, 04 Sep 2010 15:20:06 GMT

I thought all it did was to create symlinks with the proper prefixes in a automated way for a given script, something that one could do manually... but apparently it does something more.

But perhaps I just did something wrong when trying to do it manually, like using the wrong runlevel folders. Or perhaps it was just the difference from priority 20 to 19 that made the difference even though I thought the key was to be a higher priority (i.e., a lower number than) than 22.

Anyway, update-rc.d is easier.

If condition "if condition ... if condition"

the dsc — Sat, 28 Aug 2010 23:26:47 GMT

If you want to have a script that would do something if either one of two consequences is true, a simple way to do that would be:

Code:

if [ condition 1 ] || [ condition 2 ] ; then
if [ condition 1 ] ; then
consequences for condition 1
fi
if [ condition 2 ] ; then
consequences for condition 2
fi
consequences of either one, but happens only once, not twice
fi
...

You check if either one is true, when "||" works as "or", then go on checking each one independently again.

But alternatively you can concatenate "checks" within "checks":

Code:

if [ `
if [ condition 1 ] ; then
consequences for condition 1
fi
if [ condition 2 ] ; then
consequences for condition 2
fi ` ] ; then
consequences of either one, but happens only once, not twice
fi
...

I don't know if it would work in every circumstance, but it worked on a simple script I've made just to test that. It's possibly not absurdly superior for that use specifically, but perhaps the principle can be a key for an easier solution in a more complex scenario. Or maybe not. There it is, anyway.

Batch-fixing symlink targets

the dsc — Sat, 28 Aug 2010 23:08:15 GMT

Sometimes you move the folder to which lots of symlinks were pointing, or something to that effect, and you want to simply "edit" these symlinks rather than recreating them pointing the correct new location. That's how it can be done with one line:

Code:

find /place-where-the-null-symlinks are/ -type l | while read nullsymlink ; 
do wrongpath=$(readlink "$nullsymlink") ; 
right=$(echo "$wrongpath" | sed s'|/whatever/is/the/old-wrong/path|/correct-new/path|') ;
ln -fs "$right" "$nullsymlink" ; done

Of course, that will work only if the files the symlinks were pointing to were all moved together to some other location, preserving the folder structure after "/whatever/is/the/old-wrong/path" on the "/correct-new/path".

Sound on flash and everywhere else at the same time, in Arch

the dsc — Fri, 13 Aug 2010 17:57:35 GMT

There had been a weird problem with sound on Arch. Whenever there was a flash item being displayed (or perhaps flash using sound, I suppose), there would be no sound elsewhere. And if there were sound elsewhere, whenever you play something on flash, it would conversely be mute.

The fix is quite simple:

Quote:

Originally Posted by Wilco

After having some troubles with alsa I managed to fix this once and for all. The problem was I could not run flash+firefox and some other application that uses sound, like xmms or mplayer.
After investigating this problem on my old laptop, searching this forum and the wiki I finally found the final solution which was not described in the other topics, so here it is:
Add to /etc/rc.conf:
MODULES=(... !snd-pcm-oss)
(don't add the dots ofcourse)
The problem with flash is that it defaults to oss if it can find oss. So no matter if you have alsa, flash will prefer oss and oss will block /dev/dsp and no other program can play any sound. If there is no oss device available, flash will use alsa and all problems are gone. No need for a dmix setup or configuring other programs, just disable the old oss system.
Hope this helps someone!

https://bbs.archlinux.org/viewtopic....532805#p532805

Helped me.

Notes on how I may have managed to get wacom tablet to work on Arch, before I completely forget it

the dsc — Fri, 13 Aug 2010 14:35:38 GMT

By "work" I mean having "hovering" and "keepshape" mode, not just a worthless slow "pen-mouse". My thinking is that it wasn't working due to concurrent configuration files. There isn't only xorg.conf, but on /etc/X11 there is a xorg.conf.d folder, and there there was a ##-wacom.conf or something, with different settings than the ones I had set manually on xorg.conf.

HAL is deprecated, or something like that, and I don't have any .fdi file for wacom in the "policy" folder.

I have a tablet-related udev rule, however:

KERNEL=="event*", SYSFS{idVendor}=="056a", NAME="input/%k", SYMLINK="input/wacom"

with the filename 10-local-rules. I don't know if it's doing anything though (I've searched for "wacom" on Debian's udev rules files, and found anything). Sometime I got to test this sort of thing carefully, so I know exactly what makes it work.

Canon Canoscan Lide 25 on Arch Linux 2.6.34

the dsc — Fri, 13 Aug 2010 14:25:55 GMT

Apparently I got it working. It did do a preview scan, with color, on xsane, as a normal user, so it's almost 100% sure it is OK. The only thing I think I may have had to do different from Debian was to set a udev rule and change user permissions/ownership on /var/lock/sane. Some people said that just udev or permissions alone would do it, but I had the impression I needed both, but that's possibly wrong. I don't see anything wrong in having both, so I'll just leave it as it is, as it is working.

References:

https://bbs.archlinux.org/viewtopic.php?pid=637941

http://wiki.archlinux.org/index.php/USB_Scanner_Support

It didn't work so readily as I expected it would, though. For some reason sometimes it fails to be detected by the scanner program (running either as a normal user or root), even though it's listed by lsusb and even (I guess) sane-find-scanner.

First it seemed to be working, but the lights lighted up, but just stood there. I tried to unplug and plug the USB cable again (probably there is something smarter and more correct than that to do, like restarting udev, I don't know). Then it wasn't detected anymore, even plugging on different slots (if that makes any difference... from past experiences I had the impression it does).

I don't know what caused it. I booted on debian and it was working. Then I rebooted on arch, and it was working as well.

Random song constrained by keyword on MPC

the dsc — Thu, 12 Aug 2010 05:17:43 GMT

Save it as a script, perhaps even as an alias:

Code:

mpc play `shuf -e $(mpc playlist -f "[(%position%) %artist% %title% %file%]" |egrep $1 | grep -Eo "\(.*\)"| sed 's:).*::g' | sed 's:(::g') -n 1`

Then type at some terminal:

mpcr[andom]

(or however you chose to name/alias it)

And mpc will play a random song from the pool resulting from the keyword filter.

I guess the whole chain of pipes could probably be elegantly replaced by some awesome sed parameter, but that will do it as it is, anyway.