From: Junio C Hamano Date: Fri, 27 Jul 2012 22:52:43 GMT Subject: Re: [PATCH 4/4] Move initialization of Git::SVN variables into Git::SVN. Message-ID: <7vwr1oyfyc.fsf@alter.siamese.dyndns.org> In-Reply-To: <20120727221924.GA8700@dcvr.yhbt.net> Eric Wong writes: > Eric Wong wrote: >> So I'll hold off until we can fix the build regressions (working on it >> now) > > OK, all fixed, all I needed was this (squashed in): > > --- a/perl/Makefile > +++ b/perl/Makefile > @@ -22,6 +22,8 @@ clean: > $(RM) $(makfile).old > $(RM) PM.stamp > > +$(makfile): PM.stamp > + > ifdef NO_PERL_MAKEMAKER > instdir_SQ = $(subst ','\'',$(prefix)/lib) Another thing I noticed but didn't say was that the top-level Makefile seems to think without NO_PERL the way to regenerate perl/perl.mak is to run perl/Makefile.PL, which is not true if the build is done with NO_PERL_MAKEMAKER. I do not offhand know why we even need to have dependency on perl/perl.mak in the toplevel Makefile (other than "otherwise nobody descends into perl/ and run make in it", which is a bogus reason---there should be a rule to run "$(MAKE) -C perl/ $@" when doing "make all" at the top-level if that is the case), but I think at least the duplicated rule in the toplevel Makefile should read something like: perl/perl.mak: ... (the dependencies) ... $(QUIET_SUBDIR0)perl ... (make variables) ... perl.mak so that the real knowledge of how to rebuild it (with or without NO_PERL_MAKEMAKER) should be in perl/Makefile.