Path: utzoo!mnetor!tmsoft!torsqnt!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!tut.cis.ohio-state.edu!ucsd!usc!snorkelwacker!bloom-beacon!LARRY.MCRCIM.MCGILL.EDU!mouse
From: mouse@LARRY.MCRCIM.MCGILL.EDU (der Mouse)
Newsgroups: comp.windows.x
Subject: Re:  Is X11R3 re-entrant?
Message-ID: <9007231054.AA02140@Larry.McRCIM.McGill.EDU>
Date: 23 Jul 90 10:54:57 GMT
Sender: daemon@athena.mit.edu (Mr Background)
Organization: The Internet
Lines: 31

> Can anyone tell me how reentrant the code in R3 is?

The stuff from MIT, run on anything approximating a typical UNIX
system, not very.

> In particular, what I'm trying to find out is "how safe it is to use
> Xt and Xlib calls from inside a signal handler (namely SIGALRM -
> alarm (2))?"

Xlib should be OK provided you use a separate connection to the
display.  Xt would be too, except that Xt does not behave well in the
presence of two connections in the same program (or so I understand).
If you use the same connection in the main line and at signal-handler
level, eventually a signal will interrupt a main-line call at a
critical time and you'll get burned.

> My application uses Xaw to set up an interactive control pannel and
> then periodically goes off, does some work and then updates output
> windows (which contain widgets) and so forth.  This "going off" is
> triggered by a SIGALRM.  So far I haven't had any problems making
> toolkit calls from within the handler, but a friend suggested this
> might be a dangerous thing to do.

You make it sound as though the main-line is not doing any X
interaction when the signal happens.  If you can guarantee this, you
should be OK.  If this is not true, well, read my paragraph above....

					der Mouse

			old: mcgill-vision!mouse
			new: mouse@larry.mcrcim.mcgill.edu