Path: utzoo!utgpu!news-server.csri.toronto.edu!orasis.vis.toronto.edu!tjhorton
Newsgroups: comp.windows.x.motif
From: tjhorton@vis.toronto.edu ("Timothy J. Horton")
Subject: Re: Timer Interrupt
Message-ID: <91Jan27.013455est.8650@orasis.vis.toronto.edu>
Organization: Department of Computer Science, University of Toronto
References: <9101241425.AA23877@alex.CSS.GOV>
Distribution: inet
Date: 27 Jan 91 06:35:12 GMT
Lines: 31

>>I am trying to make a main event handler which will run for at least
>>the given time, and returns the control back to the program.
>>That is, it should be a temparary XtAppMainLoop.
>>I made up the following function:
>>
>>void TempEventH(XtAppContext app, unsigned int time)
>>{
>>  XEvent event;
>>  XtAppAddTimeOut(app, time, NULL, NULL);  /* put up a timer event */
>>  while(!(XtAppPending(app) & XtIMTimer)){ /* until it arrives */
>>    XtAppNextEvent(app, &event);           /* do the event  processing */
>>    XtDispatchEvent(&event);
>>  }
>>}  

I might speak from ignorance, but I have the distinct impression (for
many convoluted reasons) that timeouts are not handled how you may think.

First, they are not an X event type (there is no "timeout" event down
in real X, to my knowledget).  I think rather that they must be dealt with
by the mainloop, involving timeouts on selects and dispatching and so forth.
My guess is that it is implimentation dependant, too.

(It can't be unix alarm timeouts, either.  For instance, System V unix
doesn't handle timer subdivisions less than 1 second, whereas BSD unix does,
so how would you get the subsecond accuracy in an X application on system V
unless you user a special device driver or something -- which would be nearly
ridiculous and I almost know they don't do it).

I'd be interested in getting the real dope on how XtAppAddTimeOut type
timeouts are (usually) implimentated.  Please, anyone know?