|
From: Dennis L. <pla...@in...> - 2005-05-12 12:34:25
|
Am Donnerstag, den 12.05.2005, 12:03 +0100 schrieb Julian Seward: > Can you be clear about what problem in Valgrind this shows up? > On the face of it, you ran a C++ program on Valgrind, and it reported > a memory leak. Well, that's not unusual. As far as I understood him, his question was about why there is still reachable allocated memory, though he tried to force gcc stl allocator to use new instead of memory pools. In 3.x version you could force it with GLIBCPP_FORCE_NEW (as stated by valgrind faq 4.3) to use new instead, so usually all those still reachable messages go away. Now he was wondering why this does not happen any more with gcc 4.x. Its simply because the now use GLIBCXX_FORCE_NEW environment variable. I think the FAQ 4.3 of valgrind should be updated to reflect this change in latest gcc compiler greets Dennis |