[bf1942] you should all know by now...

[ScratchMonkey]

--On Saturday, February 22, 2003 10:20 AM -0500 "Ryan C. Gordon" 
<icculus at clutteredmind.org> wrote:

> No...it's bad luck in most cases. Doesn't seem to follow a pattern. That's
> how these things almost always are.

Drat! ;)

> I did spend some time trying to get the server to build with Intel's C++
> compiler, too...sometimes that helps to expose bugs.

Yep, I also like to use two compiler's differing views of the same code to 
reveal bugs. The main app I work on runs on 80186 and an Analog Devices 
DSP, so it not only sees two compilers, but two drastically different ones 
(Borland and ADI). It's in the process of being ported to yet another ADI 
CPU. The 3 architectures are radically different from each other (the DSP 
has 32-bit chars, ints, and longs) so this really reveals the architectural 
assumptions. For instance, casting to char does NOT mask to 8 bits on the 
DSP, so I defined a MASK_TO_8_BITS macro to handle that.