On Mon, Feb 22, 2010 at 4:05 PM, Eric Blake <ebl...@redhat.com> wrote: > According to Michael Goffioul on 2/21/2010 4:09 PM: >> I think that because of the outer "if test $ac_cv_func_gettimeofday = >> yes; then", >> REPLACE_GETTIMEOFDAY can only be set to 1 on systems that already >> provide an implementation of it. > > Correct. But a platform that lacks gettimeofday altogether should also > compile the replacement file (but with REPLACE_GETTIMEOFDAY left alone). > >> Is this a bug? If not, what is supposed to happen on system that do not >> provide gettimeofday? > > An inadvertent regression, particularly since the documentation already > mentions that mingw lacks the function. I will commit this after testing > a cross-compile to mingw; let me know if it helps your situation (I don't > use MSVC).
Thanks. I tried the patch and gettimeofday is not compiled in. However, there's still a problem as sys/time.h does declare gettimeofday (not rpl_gettimeofday), but gettimeofday.c unconditionally implements rpl_gettimeofday. So I know I'll have a problem a link stage. Michael.
