On 04/16/2014 09:25 PM, Assaf Gordon wrote: > On 04/16/2014 03:31 PM, Pádraig Brady wrote: >> On 04/16/2014 06:33 PM, Paul Eggert wrote: >>> Pádraig Brady wrote: >>>> + [case $host_os in >>>> + mingw*) gl_cv_func_expl_buggy="guessing no";; >>>> + *) gl_cv_func_expl_buggy="guessing yes";; >>>> + esac]) >>> >>> Sorry, what does this have to do with mingw? Why would we guess yes for >>> (say) glibc? >>> >>> Instead, please guess no by default, and guess yes only on OpenBSD hosts. >> >> Oh good point. >> I thought it might have some unmentioned significance on mingw >> but it's probably a copy & paste issue (from roundf.m4?), >> especially since "guessing yes" would have never matched the >> subsequent check anyway. I'll merge this in... >> > > Sorry - that was my mistake (indeed, copy&pasted from m4/stat.m4). > > If I understand correctly, that part is relevant for cross-compiling. > On second thought, I think it's better to "guess no" for all systems - > since even for OpenBSD, most versions have a functioning 'expl' (this bug has > already been fixed in their main repository, I believe) - perhaps even remove > this part all-together?
Better to use something that will work always. We can revisit in future as these systems get deprecated. I've pushed as per the last diff. thanks, Pádraig.
