From: Junio C Hamano Date: Mon, 19 Sep 2005 16:34:35 GMT Subject: Re: [PATCHES] CURLDIR support, more usable cmd-rename.sh Message-ID: <7vhdchqan8.fsf@assigned-by-dhcp.cox.net> In-Reply-To: <1127139079.781.9.camel@divert> Patrick Mauritz writes: > diff -ur git-core-0.99.7.orig//Makefile git-core-0.99.7/Makefile > --- git-core-0.99.7.orig//Makefile Mon Sep 19 12:43:04 2005 > +++ git-core-0.99.7/Makefile Mon Sep 19 12:45:45 2005 Why not a patch from either git or cogito or stgit ;-)? Anyway, thanks for your patch. > -CFLAGS = -g -O2 -Wall > +CFLAGS = -g -O2 Not relevant to the current topic as far as I can tell (unless your non-standard curl include files trip under -Wall). > +ifdef CURLDIR > +CFLAGS+=-I$(CURLDIR)/include > +endif > ALL_CFLAGS = $(CFLAGS) $(PLATFORM_DEFINES) $(DEFINES) > > prefix = $(HOME) > @@ -261,7 +264,11 @@ > git-ssh-pull: rsh.o fetch.o > git-ssh-push: rsh.o > > +ifdef CURLDIR > +git-http-fetch: LIBS += -L$(CURLDIR)/lib -R$(CURLDIR)/lib -lcurl > +else > git-http-fetch: LIBS += -lcurl > +endif I do not have any objections to helping installations where CURL is installed somewhere nonstandard, but if that is the case I am wondering if assuming that they are all in the same directory hierarchy covers enough. CURLDIR=/usr/local/ would be OK and CURLDIR=/opt/curl/ would be as well with this patch, so this probably is fine. Installations without ld.so.conf need -R, but is it safe to always force -R like this? Also I wonder if we need similar customization for openssl stuff, but that is a separate topic. > --- cmd-rename.sh~ Mon Sep 19 12:53:40 2005 > +++ git-core-0.99.7/cmd-rename.sh Mon Sep 19 12:53:58 2005 > @@ -4,7 +4,9 @@ > while read old new > do > rm -f "$d/$old" > + if [ -x "$d/$new" ]; then > ln -s "$new" "$d/$old" > + fi No questions about this part. Thanks for the fix.