Eric Sunshine <sunshine <at> sunshineco.com> writes: > > On Fri, Aug 14, 2015 at 6:22 PM, Junio C Hamano <gitster <at> pobox.com> wrote: > > Eric Sunshine <sunshine <at> sunshineco.com> writes: > > > >> On Fri, Aug 14, 2015 at 5:02 PM, Junio C Hamano <gitster <at> pobox.com> wrote: > >>> Eric Sunshine <sunshine <at> sunshineco.com> writes: > >>> > >>> I do not think that is anything new. We always have assumed "some" > >>> version of Perl available in order to run t/ scripts. > >> > >> True, but prior to 527ec39, without Perl available, git itself could > >> at least be built and used (with some commands unavailable), even if > >> it couldn't be fully tested. As of 527ec39, however, git won't even > >> build because common-cmds.h can't be generated. > > > > I wouldn't bother digging in the history myself, but I am reasonably > > sure that the current genereate-common-cmds is not the sole instance > > that we relied on Perl to build (not test) in the past, and that is > > another reason why I do not think this is anything new. > > Hmm. In my tests by setting PERL_PATH to a bogus (non-existent) > command, prior to 527ec39, git builds successfully, whereas, following > 527ec39, it does not build. But, perhaps I overlooked something...(?) It builds but there will be at least 3 commands that won't work: git-submodule git-request-pull git-am I'm considering to add perl dependency as mandatory on FreeBSD ports tree, and maybe this NO_PERL option doesn't make more sense nowadays... -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html