Andreas Sæbjørnsen wrote:
I still experience the same problems as before with regards to this memory bug. I am not able to test this with the wave driver though as it compiles with a lot of errors in a lot of header files with gcc 3.4.3 on linux, but it shows up in the lexed_tokens example.
That's puzzling me. Why the Wave driver doesn't compile for you?
Are you sure to really have the corrected version (I've replaced all memcpy's in cpp_re.cpp with memmove's, which should avoid the problem reported by valgrind)?
Seems that the public CVS currently is a bit out of sync if compared to the 'real' CVS database. At least ViewCVS shows a state which is 5 weeks old.
This would explain your compilation problems. And I don't know where you may get a current CVS version, sorry.
Ok. That explains. I will probably try to apply a change from memcpy to memmove's. It would be great if you could just send me your cpp_re.cpp on email in case there are some details I missed. If there are some newer CVS than the public boost CVS it would be great if I could manage to check out from it since I do some work with Wave where I apply it to very complex codes, so I will see if I can find out how to get this kind of read access. Do you use a different cvs or do you have a process of multiple versions of the CVS? Thanks Andreas