-=> Quoting David Noon to Phil Crown <=- 

 DN> On Thursday, 96/07/04, Phil Crown wrote to Daniel Lynes about
 DN> "sys3171" as follows:
 
 PC> I was getting this error under Borland's OWL in a secondary
 PC> thread.

 DN> Hi Phil,

 DN> There is a well known bug in the Borland compiler, not generating
 DN> stack probes to commit pages of the stack. This causes abends in
 DN> secondary threads, since the stack is not committed automatically in
 DN> these.

I had heard of this bug before, but do you know if it exists Borland
v2.0, or was it only in earlier versions?

 DN> There is a patch to the RTL available from Borland's Compuserve
 DN> forums. This commits the pages when they are first allocated, which
 DN> can be a little inefficient.

I will look on their Web page and see if I can find it.  Don't remember
seeing anything like that last time I was there, but I wasn't really
looking for it.  Thanks.
 
 PC> I've switched to VAC++ Trial Copy (from Devcon) and am not getting
 PC> any errors. :-)

 DN> That is a most sensible fix. You might like to make it permanent.

I'd like to, but VAC++ is slow on my machine, a 486/66 with 20 megs of
RAM.   I'd expect it to be somewhat slow, but even something like
right-click the mouse to bring up the context menu is slow!  Its as if I
am running OS/2 with pmshell.exe in 8 megs, if not slower. I may try
going up to 32 megs (if memory prices keep falling :-) and hope that
will help.

Now, I'm back to BC++/2 and using OCL v1.20, a C++ Class Library for
OS/2 which supports all the major compilers and comes with complete
source code and is free.

Phil - Fido:  1:124/8014.0 | Internet:  phil.crown{at}bluecafe.com

--- Blue Wave/OS2 v2.30