nanogui: GrPeekEvent() causes client to crash
Subject:
Re: [nanogui] GrPeekEvent() causes client to crash
From:
"Greg Haerr" ####@####.####
Date:
13 Jun 2001 04:54:21 -0000
Message-Id: <004001c0f3c4$a3864840$3aba46a6@xmission.com>
: The problem appears to be related to how fast the app runs (i.e. how
frequently it calls nano-x). Might be some sort of overrun bug??
Yes, we fixed a client-side event message overrun bug. The
fix resulted in the creation of a specialized client-side event
queue, and routines in src/nanox/client.c like GetNextQueuedEvent
and QueueEvent. I can't exactly remember the bug, but I think
it had to do with an error that occured only if the client program
slowed down enough that an event message sent by the server
would be still on the wire, not read in, and the client would make
another server call that would result in the event message being
lost. That would then cause the client/server to become unsynchronized.
I think your fix may be that the same fix needs to be applied in
the PeekEvent case. Check out client.c and look for the *Queue()
function calls, hopefully this helps.
Regards,
Greg
I'm gonna have a go at tracking it down, could you tell me more about
the bug that you guys fixed in pre6? (i'll look at the changelog but
mostly it isnt detailed enough...).
Oh and I actually meant GrPeekEvent()/GrGetNextEvent() combination
instead of GrPeek/GrCheck...sorry for the confusion.
-----Original Message-----
From: Greg Haerr
Sent: Fri 6/8/2001 9:26 PM
To: Constantin Pavlicu; ####@####.####
Cc:
Subject: Re: [nanogui] GrPeekEvent() causes client to crash
: I'm trying to use GrPeekEvent() to check for events before calling
GrCheckNextEvent(). Unfortunately adding a call to GrPeekEvent() causes
the nanox client to crash with:
nxclient 1497: Wrong packet type XXXX (expected 22)
-- What version are you running? I know we fixed a client/server
problem in pre6b that was similar to this.
Regards,
Greg
