Bug#606618: glib2.0: g_timeout_add() cannot handle very large timeouts

[Jakub Wilk]

clone 606618 -1
severity -1 minor
reassign -1 syslog-ng 3.1.3-1
retitle -1 syslog-ng should use g_timeout_add_seconds (rather than g_timeout_add) where possible
thanks

* Emilio Pozuelo Monfort <pochu at debian.org>, 2011-01-04, 10:51:
>>>I have "stats_freq(86400)" in my syslog-ng.conf, and it used to work 
>>>fine until I upgraded libglib2.0-0 from 2.24.2-1 to 2.27.4-1. 
>>>Apparently statistics are now logged every 8m30s (rather than every 
>>>24h, as I requested).
>>
>>I turns out this is a bug in glib. The attached minimal test program 
>>should go ping every ~24 hours, and it apparently does with glib 
>>2.24.2-1, but not with 2.27.4-1:
>
>Any reason you don't use g_timeout_add_seconds for large timeouts?

I was just copycatting what syslong-ng did. I can reproduce this bug 
with g_timeout_add_seconds, too.

>that will cause less wakeups (so less power consumption) and I guess 
>precision is not a concern for such timeouts...

Indeed, sounds like a good idea.

-- 
Jakub Wilk