$MY-CROSS-COMPILE-PREFIX/mingw32/include/time.h does not have localtime_r nor gmtime_r. tor_localtime_r and a procedure for creating useable versions suitable for win32 is here: http://archives.seul.org/or/cvs/Feb-2005/msg00222.html
Created attachment 23254 [details] yuk! do not try this at home! badly-hacked get-it-to-compile jobbie.
When will this patch go in, needed for compiling on mingw32. I'm cross compiling on mingw32 on Fedora 10 and this is needed and it works.
A patch originally described as "badly-hacked get-it-to-compile jobbie" needs to be reworked to have a chance of going in.
(In reply to comment #3) > A patch originally described as "badly-hacked get-it-to-compile jobbie" needs > to be reworked to have a chance of going in. :) all it needs is someone else to review it, and come up with either a yes/no or an alternative. for example, why does Win32 native build succeed (or does it?) does MSVC have localtime_r does MingW32 _native_ have localtime_r ? so - should this be #ifndef MINGW32 (or equiv) rather than #ifndef WIN32? is the non-thread-safe version of localtime and gmtime acceptable to use? should there be thread-locking around the use of localtime and gmtime? is memcpy acceptable? empirical evidence suggests that "yes" is the correct answer to the last three of these questions.