> 1. Borland C++ for OS/2 is the buggiest C compiler ever devised,

ac> I use Borland C++ for OS/2 1.0.  I haven't come across any bugs, but then
ac> I haven't pushed it very far.  Are there any bugs I should be aware of?

From the top of my head, 

1. setvbuf with values greater than 64k fail.

2. The example program in the Borland book on how to use signal()
doesn't compile, and obviously when you're trying to use it in 
real life (e.g. trying to recompile zip/unzip I think it was), it
fails.

3. Parameter "-4" causes invalid code to be generated.

Don't bother upgrading to BCOS2 1.5 though.  Next time you FREQ
something off me, pick up MYST*.*, which documents all the 
mysterious compiler bugs I have.  I started this list after I
already had BCOS2 1.5 so I didn't document the 1.0 bugs here
(mind you, I can probably retrieve them all from this echo, as
I posted every one in full here, and I have a 7000-message
deep (and growing) messagebase.  BFN.  Paul.

MYST2.ZIP - CSET++/2 2.0 causing linker warning
MYST3.ZIP - BCOS2 1.5 bug
MYST4.ZIP - BCOS2 1.5 bug
MYST5.ZIP - BCOS2 1.5 bug
MYST6.ZIP - Watcom C++ 10.0a bug
MYST9.ZIP - Problem linking with BCOS2 1.5 + OS/2 2.1 doscalls.lib
MYST10.ZIP - BCOS2 1.5 TLINK internal bug
MYST11.ZIP - BCOS2 1.5 TLINK internal bug
MYST12.ZIP - Problem using BCOS2 and LINK386
MYST13.ZIP - Watcom C++ 10.0a bug
MYST14.ZIP - Watcom C++ 10.0a bug
MYST15.ZIP - Watcom C++ 10.0a bug
MYST16.ZIP - GNU/EMX 0.9a gasp bug
@EOT:

---