26 Oct
2006
26 Oct
'06
9:41 p.m.
Hi Christian, Christian Henning wrote:
Hi Christopher, I think we can drop that issue for now. I cannot find my MFC program that produced the leaks. I'm sure that if there is a problem with MFC's leak detector it will come up again. Sorry for the noise.
Well, I did manage to reproduce a leak report, and it always seemed to be related to an operation started in image_quality_source::do_write(). In that function you're sending data from a std::string local variable called "msg". This is wrong since you need to ensure that the data being sent is valid until the asynchronous write's callback handler is called, so perhaps it should be using a class data member instead of a local variable. Cheers, Chris