I decided to do some house cleaning and close some bugs tonight.... I ran into a bit of a stumper on bug #26021 for xfstt. The bug arises due to a slight oversite on my part (I wrote the pid file making/removing code...which the upstream author happily accepted). There is one case where xfstt will exit BEFORE its main loop and after the pid file is made. This is a problem because since xfstt is a deamon by its nature, it generally only dies by a signal...so the code to remove the pid file was in a signal handler. The rest of the bug is a init.d script problem...the script was something I concoted to do the job...however is lacking. I am cinsidering switching to start-stop-deamon but I have a few problems... Here are the important issues... xfstt fork()s to service clients (xservers generally). Thi sallows it to service multiple connections. The original parent xfstt can be killed and restarted painlessly without effecting the children who are currently serving connections. This IS important. The reason is simple. Many Xservers (I kno wmine does) die miserably when fontpaths unexpectedly dissapear. Just start x and killall xfstt. ... Most Xservers which I have tried will crash the very next time they try to acess the fontpath. So thus here is the problem: I would like the init.d script to PROPERLY check that a pid file already exists and is a VALID pid file (ala start-stop-deamon) but I DO NOT want it killing all the xfstt "children" or refusing to start because of those children any ideas? -Steve -- /* -- Stephen Carpenter <sjc@delphi.com> --- <sjc@debian.org>------------ */ E-mail "Bumper Stickers": "A FREE America or a Drug-Free America: You can't have both!" "honk if you Love Linux"
Attachment:
pgpXk0TKjqcjH.pgp
Description: PGP signature