Until now every thread pool always had at least one tread waiting to avoid
authorSebastian Wilhelmi <wilhelmi@ira.uka.de>
Tue, 17 Apr 2001 11:48:45 +0000 (11:48 +0000)
committerSebastian Wilhelmi <wilhelmi@src.gnome.org>
Tue, 17 Apr 2001 11:48:45 +0000 (11:48 +0000)
commit313ed5dc19898f7d2c868c828443bdf2227eb1a5
tree42ee7e1f477e6a5dee888052b0552e5c4944fe26
parent525689823d7dcd269b052c58545a5ff1d8db65f8
Until now every thread pool always had at least one tread waiting to avoid

2001-04-17  Sebastian Wilhelmi  <wilhelmi@ira.uka.de>

* gthreadpool.c (g_thread_pool_thread_proxy): Until now every
thread pool always had at least one tread waiting to avoid
switching overhead in case a new task would be added soon after
one finished. This however means a big waste of threads, if many
mostly inactive thread pools are involved. Now such a waiting
thread will only wait for half a second (This value is of course
very randomly picked) and go to the global threadpool afterwards.

MCVS: ----------------------------------------------------------------------
ChangeLog
ChangeLog.pre-2-0
ChangeLog.pre-2-10
ChangeLog.pre-2-12
ChangeLog.pre-2-2
ChangeLog.pre-2-4
ChangeLog.pre-2-6
ChangeLog.pre-2-8
glib/gthreadpool.c
gthreadpool.c