Concerning bugginess (lack thereof) and upgrade temptation:
I'm currently writing software for a wireless RFID probe and
base station receiver for the laboratory market. The probe and base both
use an ATmega128. Both sets of software (probe and base combined) are
currently over 15,000 lines of code and I'm not even done yet. Before this
project I was using the IAR compiler for the Atmel AVR on othe projects.
Due to cost considerations in "upgrading" their product to use the
ATmega128 I decided to finally check out GCC. At the time, they were just
starting to integrate support for that product. I downloaded the avrgcc128
beta 2 version and I've been using it ever since without any problems
whatsoever. I'm not upgrading the compiler during this version unless I
have to. I probably will upgrade in between versions of this product, but
I don't expect any real issues to come up. The support via the list has
been great, and you really get timely information about any problems with
the compiler / library, which may or may not affect you. The only down
side (and its a small down side) is that documentation is not exactly
"centralized". If you want an answer, you can find it in some
documentation somewhere; you just have to be willing to search around and
follow different links to find it, but it can be done.
Eric
-------Original Message-------
Date: Wednesday, July
31, 2002 11:57:11 PM
Subject: Re:
[avr-gcc-list] avr-gcc credibility
<snip>
The trap with gcc (and avrgccc) is that it's tempting to
upgrade every time there is a new patch. Don't do that unless there is
a compelling reason to. New patches introduce new bugs and new
behaviors.
Test the compiler, examine it's output and build up your
confidence this way. So far on this forum there have been very few real
bug reports. That's rare.
Sander
|