On 7 Sep 2005 at 13:02, Joerg Wunsch wrote:
My own experience says that it's best to get used to the artefacts
caused by debugging optimized code, as a good number of programming
errors will only be found once the optimizer is turned on (as the
compiler will then aggressively impose all its rules onto the source
program).
yes, that is my experience, too - under IAR I even had the problem
that some errors only came up with debugging info turned off... :-(
But still, I think I know the artefacts in debugging optimized code,
but at the moment, the combination WinAVR <-> AVRstudio is unusable
for debugging. At least I can not correlate the steps, breakpoints
and variable accesses to anything even remotely near the point where
AVRstudio thinks the ATMega is stepping right now.