Parent Directory
| 
Revision Log
| Sticky Revision: |
fixing the debug deletion (previous commit)
- fix the debug log which did not store the log that was gotten before teh disk was writtable - fix the debug log so that it inherits parent shell's log buffers - other minor stuff
- fixed a bug with lgo and debug in splashy-init (they would simpyl not work) - fixed a minor bug in the -S option for debug() - now the log gets the output from splashy_update - Added the EXTERN_PROGRESS option to be compatible with debian's forthcoming usplash_write support - Added support to run some commands (POST_EXEC_SERVICES) to hopefully fic the unicode_* problem - few minor corrections - correctly fixed the walkaround by Lunar^ - Now removed the poid at boot before running splashy, as Luis asked - corrected the scripts README and minor update to the splashy README
- minor bug fix by Lunar^. allows splashy to be started from rcS.d when not in initramfs mode
- fixing splashy-init's error messages
- Correctly enter newlines in splashy-init's log - Prepare for sending --debug to splashy, when it supports it, from splashy-init (in debug mode only) - splashy-init seems to have now completly replaced the old splashy init scripts. So the README file no longer talks about it as a transition script - splashy-init will now not show that "already running" warning during bootup and inter-runlevel-switching. This was causing confusion in users - Fixed the LSB initscripts comments
again compatibility with dash: read must be given an argument in non-bash shells
too much javascript
remove redundant code. dd's status=noxfer is too recent an option to include it here, since many systems could not have it yet
fixes the command cut not being un /bin but in /usr/bin, thus being unusable in the script. Also fixes END_MATCHES's default value which was broken and slightly changes the way it is used so that it is more comfortable
- new test package - removes debug text - clean debug messages
update for current LSB 3.1 RC3 compatibility
i retract from my last commit: the whole idea of the shutdown is to make sure splashy-init is called from within the shutdown rc script so that it never gets killed by killall5, so it is nonsense to pretend it will be killed during shutdown
fixes the untested and experimental code in previous commit. fixes halt/reboot-not-going-to-100% issue
typo in last commit. also more debug
More and MUCH better debug information (includes PID in header to distinguish concurrently running instances, and signals traps inform debug now). absolutely user-hidden log (no more outputs about errors on write). almost all output can be hidden now even when /dev/null doesn't exist (experimental and untested)
urgent yet untested fix: the framebuffer+console devices check had been broken. this is just a partial fix. also log is now buffered until the filesystem is writtable so no log is lost
debug message to detect an unexpected exit
otavio suggested to loop a maximum of times to check for the framebuffer device and exit if failed, instead of idling forever, for those using udev but not havin FB configured. so here is the answer ;-)
log must only be written in splashy-init when on debug-mode
fixing splashy-init on reboot/halt
splashy-init support for log
restoring old splashy-init script as suggested by lemsx1. Merged lemsx1's changes. Backported to the progress-is-absolute approach. Updated the /etc/defaults/splashy file
- turns progressbar on
- puts back the code to switch to X tty
- makes sure splashy is dead dead dead after the daemon_client exits - code cleanups. be more polite when killing splashy - splashy init old code
- better detection for splashy running
- fixes minor bugs
- changes TTY var to CHVT_TTY to be consistent with splashy-init - more code cleanups. copied over functions from older version of the script - we only need rcS.d/S02splashy and rc[06].d/K01splashy
- code cleanups
- call yourself again in daemon mode
- get list of scripts for progressbar
- trap only needed signals
- needs get_scripts* functions
- more bug fixes
- minor fix
- daemon client code in
- changed variables to new names for new splashy-init. more debian-like names - skeleton splashy-init. need the client code that runs in daemon like fashion
- test package - put back progress to exact percentage - attempts to fix splashy-init bugs - comment fixes
First try to do not restart splashy when it is already running.
- new test package - we always use fifo now - set perms for fifo - let splashy deal with its own fifo in initramfs - no need to kill splashy now
fixing script from exitting when fifo failed instead of retrying. also added all the necessary debug information for tracing progress problems with the increment progress method
added some debug. fixed a bug where repeating progress was sent to splashy. disabled the restore-bar-on-reset feature, since it cannot be used with current progress model. DEBUG=2 is now working. updated readme file
yet another fix regarding the same start-once-completed bug. also changed failure on stop to a warning instead of an error
correctly fixed the start-once-completed bug supposedly fixed in previous commit. merged two little pieces of code into a more consistent and readable way. added a DEBUG=2 option which will not only display debug information but also disable splashy form starting and act as if it was running
fixing a bug when splashy-init is launched when all scripts have been already executed. and some cosmetics
progress fixed and tested to work. Also enablling all debug again. fixed help info. fixed the check for /dev/vcs1. fixed the initial startup check
Untested, but fixes any things. going for qemu noq. otavio, same with you ;-)
didn't i clearly say 'nothing to see here'?
Some more debug. Dedicated to otavio
nothing to see here. move on! :-*
NOTE: THIS IS UUUNNNTTTEEESSSTTTEEEDDD (UNTESTED) CODE. Committing just for the sake of testing and because I'm going to sleep (and I dont' rely on my dying HD) - Updated script README file - Check for /etc/splashy/config.xml - Fixed the layout back to "my style" (80-char lines, 8-chars-tab-idented) - splashy_update is now set froma variable - Disabled INVERSE_DOWN since it is unusable on server side and the - Disabled FIFO_FILE and all of its checks since splashy_update now does all direct use of it and the script doesn't anymore - Some code clean-up - Recovered the code deleted in the last two months and commented it out. It might be useful in the feature so it is better to have it commented than to jsut not have it - get_X_tty() now returns 0 when it detected an X tty or 1 when it didn't - Only update on completion if it wasn't already in 100% - Moved "send_command allowchvt" inside the if condition (there are cases in which the vt will not be changed) - Now $CHVT_TTY is never modified (it is a value read from config, so itt should be treated as read-only) - Commented out the old fb-detection (which almost always worked afaik, but was very expensive and hackish) with the simple "if /proc/fb exists and contains data, fb is usable" - Fixed the progress detection routine. Now it calculates the percentage in the old way and keeps the old percentage in a variable and then extracts the latter one from the first one to know the difference and send it to splashy server - Also changed the output "Division by 0" with a much more user friendly "No scripts detected!" - Removed a floating fi which finished no if! - Moved Luis' comment about the new "send only the difference progress" technique to the place where this really matters - There is no need to get all `pidof splashy`s and then just get one. `pifod -s`does that for us
- adds common_macros - code cleanups - allow changing of vt's earlier
- break after killing splashy
- new test package with splashy_grep - minor bug fix
- repository cleanups - force kill splashy after initramfs splashy runs - no signal trapping for now
- turn debuggin off
- minor bug fix
- fixes bashism
- attempts to handle USR signals
- use Jacobo's calculations again. minus the xscript number since we don't want to do this now - code cleanups - ignore bz2 dist file - test package - avoid warning about missing new-line at end-of-file - fixes minor bug in initramfs when copying fifo
- fallback to tty1 if we are still in auto past the X test
- quoted keyword exit
- sends one string to splashy update
- use fifo and pid as defined in /etc/splashy/config.xml - needs to use strings in post* scripts -x checks - no more oldprctg
- minor touches
- fixes syntax error
- exit if command worked - add comments and uses full path to splashy_update
- splashy_update doesn't need other libraries - new path for splashy_config
Try 10 times and go to hell.
- adds initramfs to dist - splashy keeps track of its percentage now - splashy update doesn't need libglib at all
- preliminary initramfs support - adds dependency on initramfs-tools - adds notes - initial initramfs scripts and hooks - adds font to install process - fixes minor bug - copy the support files as well to initramfs - do not exit if we can't read the fifo - be very unkind towards splashy when PRELEVEL=N - use kill and pidof instead of killall
- test package - fixes typo
- debug is off
nothing really, just a tiny check to make user-launching easier in some special case. this also refreshes project's alioth stats ;-P and serves me to test kdesvn hehe
fixed header :S. inter runlevel switch is working and tested. Removed non-script-related comments. fixed the way it got PREVLEVEL variable. updated readme to reflect a new reality hehe
better way to check if the system is booting
fixes the case when manually going into runlevel S
ups
Added values false, FALSE, False to is_enabled(). redesigned the runlevel detection and usage by redefining the meanings of RUNLEVEL and PREVLEVEL along with target_rl and target_rl_was_set. sanity cleaning such as making local all trivial variabels and commenting exactly what each function returns andthe variabels it exports. some other minor fixes or cleanups. fixed a major bug in previous commit with is_client_running() broken. fixed it also so it handles a sudden runlevel S during a non-S runlevel. removed deprecated (mostly perl) code
added a note to the Readme file. fixed a bug in previous commit (which broke the script). fixed the succeeded counts which was broken in previous commit. fixed the inter-runlevels-switch detection, since it was breaking non-boot/down runlevel switches. this script is tested here now, of course, with dash
now supports progressbaron* correctly. untested
support for hot runlevel switch to S. also in Readme.txt
tested with X and it works. fixed a bug in the way it checked the existance of files. fixed a bug in hanlding the init stop calls. fixed a big bug in the way it handled post-S runlevel and while fixing it, cleaned that part of code a lot. fixed a bug in a recent commit about a variable being tested as integer when it wasnt yet set. Fixed the readme installation section and added an uninstallation section
Removed comments to make them into a Readme file, still needs to remove the non-script-related TODO stuff. Fixed a little very bug in is_enabled(). support for progressonbar disabled. Fixed a serious bug where config_parse() would fail to do its job correctly in some cases. Fixed all error messages to display correctly. fixed xml parsing. fixed config reading, which would report error when not existing. added support for ^ and $ in END_MATCHES. fixed level setting in END_MATCHES. added CHVT_TTY values "auto" and "none". Added support for FRAMEBUFFER device setting. design change: is_client_running doesnt require arguments. fixed bug in perform_complete() where it would not detect correctly if the progress has to be sent or not. Now start_server() does all th server starting job and accepts an argument, -s, to display errors or not. added more checks to server launching and made them more user-friendly. changed the way the script handles its calling execution (read the readme in next commit). now it idles at startu when the fb device is not yet created but udev will run. now it doesnt get the vga= value from /proc/cmd but from env. redefined signals to handle modes better (read readme in next commit). get_status(), progres2percentage() and wmain while loop redesigned to be much clearer now :). redefined the init stp mode to use stop mode silently. and the usual variables renamed, and code readability thingies. probably some more fixes btw
now supports a FRAMEBUFFER config variabel to set the frambuffer device to use (standard X token). when shutting down, do not exit splashy, so that the image is left there hanging (this should eventaully be used to put a 'you can now power off the computer' image). fixed a bug where splashy would exit when detecting the script before gdm,xdm,... instead of when reaching the script. silenced the server so that all errors are only reported by splashy, which already should do quite good at that task. fixed a bug where ifa user entered verbose mode _very_ quickly when the script had been able to start splashy in 'start' mode, it would splash the imag again when entering 'daemon' mode. added the notes about sourced scripts by thebonsay at #bash. somewhat a bug fix, try to find out if it is between runlevels S and X when no rc(S) script is found ot be running. made the splashy server launching eror report much more user-friednly and precise. also fixing, on script side, splashy binnary's lilo vga detection bug. this script, right now, should be considered in beta stage, or pre-release ;P
This adds the ability to define the end by setting a number between 0 and 99 (using one or two numbers only), independently of the script name. Also adding the hint to detect lilo, which has taken me all afternoon, so I don't forget
now splashy runs from the very start if /porc is mounted, thus no text is shwn at all (except for 'splashy is running'). Also wrapped some lines to the 80 chars console limit. removed one of my J221 bug-questions thanks to lemsx1 and one og his suggestions because it was not related. Also fixed the script so that deprecated code can easily be removed before release
- adds notes in response to J221 notes
the mean feature of this patch is the removal of perl. in many ways it is now faster than with perl, though in some cases it might be slower. i'd say it is just average, but still believe it is somewhat faster. appart from that, fixed the /dev/null test to character, not block device. fixed the script so it now shuts down too. all modes work now (reload, try-restart, restart, force-reload, status, stop, ...). Removed the use of basename in favour of bourne shell variables expansions. fixed the scripts detection though still no way to detect sources scripts. added 3xdm and startx to END_MATCHES and switched its value to bourne style (instead of perl style). disabled INVERT_DOWN since splashy binary can't support it yet. removed KEEP_PROGRESS in favour of unset END_MATCHES. getscripts no longer has a 255 scripts limitation. also pid isn't written now on shutdown. fixed bug where i though 99 was the maximum progress but it is 100 (obviously). BUT, introduced a new unfixable bug: /etc/rc?.d/* filenames cannot contains spaces. please test
renaming splahy-daemon to splashy-init as suggested (and I agree) by lemsx1. if it even gets to be release, it should be THE init script, thus being an appropiate name. renaming to 'splashy' would make confusion with the binary
lemsx1's fixes + a typo
this is a new yet-to-be-approved splashy initrc script which should merge all the current (let's say 'old' from now on :P) splashy initrc scripts. it should also provide quite a bunch of new features and accurate measures, hopefully not increasing the CPU usage much. it is very unstested, not yet releasable, though usable in most cases (ehm, all that i have tested hehe). this is the last known-to-work version of the script. I am committing it so i can revert it easily if i find it is completly broken someday (since i'm not keeping backups. i'm lazy). Most of the comments in it will be removed eventually, either moved to a readme or completly removed since they are my personal agenda :D (did i said i'm lazy?) so don't scream to me before it's time! Ok, so that was the introduction to the new file. short, wasn't it?
This form allows you to request diffs between any two revisions of this file. For each of the two "sides" of the diff, enter a numeric revision.
| ViewVC Help | |
| Powered by ViewVC 1.1.5 |