<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type"> </head> <body bgcolor="#ffffff" text="#000000"> Colin Walters wrote: <blockquote cite="mid:faa16b610802210659n4cf1e62le20460667c9738c2@mail.gmail.com" type="cite"> <pre wrap="">On Thu, Feb 21, 2008 at 8:08 AM, Simon McVittie <a class="moz-txt-link-rfc2396E" href="mailto:simon.mcvittie@collabora.co.uk"><simon.mcvittie@collabora.co.uk></a> wrote: </pre> <blockquote type="cite"> <pre wrap=""> This is Havoc's rationale (which, FWIW, I don't agree with) for making libdbus exit your process on disconnections - if you have written the necessary code to handle disconnections gracefully, then you can safely disable the automatic disconnections. </pre> </blockquote> <pre wrap=""> I thought the rationale was simply that the default DBus connections to the system/session buses are tied to the lifecycle of the operating system and desktop login session, respectively. So unless your application is very special in some way, it should exit when the session or system bus gets restarted. And no OS vendor should be restarting the system or session bus on upgrades</pre> </blockquote> Looks like I made some progress : receiving such a destroy event seems related to me not calling dbus_g_thread_init(), I was calling g_thread_init instead. While I agree i did the wrong thing, there was not a single warning before this exit. I've spent one full day to find it. Can someone please confirm if an inconsistency is detected (due to multithreading) that dbus will exit? Frederic </body> </html>