RE: idle function + g_usleep = recursion?

["Freddie Unpenstein" <fredderic excite com>]

> I wonder what g_usleep does if it is called from inside an idle
> function.  Could that lead to recursion since the idle functions is
> called when the cpu has idle time?

Idle functions in Glib are basically a kind of timer, with a zero timeout, and a very low priority.  The main 
loop does its tricks with file handles (GUI events, new data to read from an internet connection, etc.), and 
then picks up the most important timeout (the one with the highest priority), that also happens to have 
expired (which is why timeouts aren't guaranteed to be called on time).  Idle functions are simply timeouts 
with a very low priority (so everything else will happen first), and a timeout of zero (so it's immediately 
qualified for execution, just as soon as the main loop is done with everything else, and finally gets around 
to looking at it).

Setting any kind of timer won't be recursion because it won't happen until the next time through the main 
loop anyhow.  So even if it was timer-based, it still wouldn't be recursion.

But then g_usleep(), like most sleep() functions, is usually of the blocking variety anyhow.  As the Glib 
documentation states, the whole thread stops for a brief period.  This will usually be a kernel call or 
perhaps even a busy loop.  But in either case, nothing can happen in the process until the sleep finishes, so 
it will never be recursive with anything (unless you could other threads executing).


(ps. If I got anything terribly wrong, please let me know! ;) )


Fredderic

_______________________________________________
Join Excite! - http://www.excite.com
The most personalized portal on the Web!