Julian Taylor a écrit , Le 17/06/2012 17:11: > how does that patch fix the issue?
Hi, The issue reported by g++ "taking address of temporary array" *is* fixed. In this context (AIUI), "temporary" means that there is no certainty that the address is still valid at the time callback_list_call_attr_1() is called. > it still takes an address of a temporary, just in a way that the > compiler can't detect it anymore. It's another kind of uncertainty you'll find waaaay more time in the navit source code. The question is then : will the array address be still in use after the exit from RenderArea::keyPressEvent() method? > a correct fix is to copy the value into the original buffer (which > then must be non-const) and not to swap the (incompatible) pointers. > or allocate a global memory buffer, copy the data in and pass that > into the callback but that might complicate freeing the memory again. A correct fix implies knowing precisely how the triggered callbacks are supposed to use the address. I don't, and will appreciate any findings you could share. Anyway, #676006 is closed and new bugs should have their own bugreports. Thanks, _g.
signature.asc
Description: OpenPGP digital signature
