Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!ra!uvaarpa!murdoch!usenet
From: djm5g@virginia.edu (David John Marotta)
Newsgroups: comp.os.os2.programmer
Subject: OS/2 message queues killing processes
Keywords: OS/2 message queues
Message-ID: <1991May21.140334.20170@murdoch.acc.Virginia.EDU>
Date: 21 May 91 14:03:34 GMT
Sender: usenet@murdoch.acc.Virginia.EDU
Organization: University of Virginia
Lines: 20
Ireallyam: djm5g

We have been having a unique problem that I thought you might have had
some experience with, and could give us some clues about its source.
Our program sends some initialization messages which synchronize the
startup of various threads within the program.  Recently the program has
been failing to startup, and the later messages never being received in
the main message queue.  Increasing the queue size from the default to
100 processed further.  Taking code out of one message and shortening the
amount of processing done in that message got us through that message
(whereas before it would die in the middle).  It would die at a given
TIME of processing, seemingly no matter what was being done.

Is there some sort of limitation on the amount of time a message can take
in order to process without being killed.  The time involved in these
messages is short (a few seconds), but when one process is done it posts
the start of the next process and there are about seven in total.  I have
only been able to think of two theories, but neither of them is
satisfactory: 1. The message queue is overflowing and important system
messages are being lost halting the system. 2. There is a time limit on
the amount of time a message can take, after this limit the processing is
killed (perhaps with a quit message? Codeview shows _a_msg_exit).