On 07/07/2025 17:51, Ramsay Jones wrote: > > > On 07/07/2025 16:40, Junio C Hamano wrote: >> Eric Sunshine <sunshine@xxxxxxxxxxxxxx> writes: >> >>>> need to link a separate library (-lsysinfo). (This would require >>>> a similar change to meson.build). >>>> >>>> - change the order of the preprocessor conditionals in the total_ram() >>>> function in 'builtin/gc.c', so that the *BSD sysctl() function >>>> (in the HAVE_BSD_SYSCTL block) takes priority over the sysinfo() >>>> function (in the HAVE_SYSINFO block). >>>> >>>> - suppress the setting of HAVE_SYSINFO when HAVE_BSD_SYSCTL has been >>>> defined (in both configure.ac and meson.build). >>>> ... >>>> The second solution would only be required by the autoconf and meson >>>> build systems, the Makefile already sets the build variables to the >>>> required values (since they are not 'auto-detected'). >>> ... >>> The final solution is almost certainly good enough (and is definitely >>> simple), although the second solution has the benefit that it "fixes" >>> the problem once and for all even if someone defines both >>> HAVE_BSD_SYSCTL and HAVE_SYSINFO (say, in config.mak), assuming I'm >>> understanding correctly. >> >> Yeah, I think I agree with this assessment. > > [Sorry for the late reply - real life keeps getting in the way!] > > Yep, I thought about including this fix *in addition to* the solution > implemented in this patch, but decided that the chances that anyone would > set both in a Makefile build was practically zero. (famous last words ;) ). > > Of course, practically zero is not zero, so we could do this in a > follow-up patch if we wanted to take a more conservative approach. > (Carlos has a series in progress which would conflict with such a Sigh, sorry, Carlo. ATB, Ramsay Jones
