Olaf.Spinczyk at informatik.uni-erlangen.de
Fri Oct 27 14:53:48 CEST 2006
Robert Scott wrote:
> On Monday 23 October 2006 21:47, Matthias Urban wrote:
>> It's still in development. There are some bugs to fix, some missing
>> semantic analyses, some features to implement. And of course a lot to
> Following on from my earlier queries, could anyone please tell me what the
> mysterious third 'm' integer parameter in CCParser::parse() does?
> If I leave it as the default 0, I get a SIGSEGV in Builder::Delete () (I can
> provide a backtrace*), if I set it to > 0, I get different behaviours, but
> all of them leave me with an empty CTranslationUnit::tree().
> Many thanks,
'm' is the preprocessing mode. If it is > 0, the code will only be
preprocessed and not parsed. Therefore, it yields no translation unit.
Please send us the backtrace and the most simple piece of code, which we
could use to reproduce the crash. Thanks.
> * - Did I mention that your weaving process makes debugging really difficult?
Well, this is always a problem with code generators. We therefore have
developed an aspect-aware debugger extension integrated into our Eclipse
plugin ACDT. If you don't want to see the generated code there, you can
simply press a button and Eclipse will hide it from you.
Please be careful with working on Puma with Eclipse. As the Puma code is
copied twice during the build process, you cannot fully exploit the
power of the ACDT.I've also never tried to debug Puma applications. The
ACDT might show the wrong files. It is one of our goals to simplify the
Puma build process.
More information about the aspectc-user