22 Apr
2002
22 Apr
'02
8:34 p.m.
It was thus said that the Great Richard Erlacher once stated:
so I meant that while you may have software that creates really pretty decent
code for one platform, that same source code may not produce as good a product
on another, and, in fact, may produce pretty bad code for yet another set of
hardware. There's plenty of evidence of that, at least according to people
with whom I routinely interact.
Depends upon the compiler, and also what you are trying to do.
I personally code in assembler for the MCU's I use
and occasionally use a
compiler if I have an environment where timing is not critical all the time
and where it's easier to use logic spelled out in 'C' or Pascal. That often
means that interrupt code is written in assembler while message generation or
communication code is produced by a compiler. Portability is seldom an issue
for me.
For embeded systems I don't think portability is an issue really. And
while I'm not sure if you work on embedded systems or not, if you are
writing interrupt code, then yes, I can see why you would use assembler for
that (heck, even the Linux kernel has the low level interrrupt routines in
assembler). Me, I haven't had need to write an interrrupt handler in oh,
maybe eight to ten years, but my code tends to be more of an application
level that may or may not run on different computers.
-spc (Have code that's moved from SGIs to Suns to IBMs to HPs to PCs
to Amigas to ... )