On 08/03/2012 10:14 AM, Stefano Lattarini wrote: > On 08/02/2012 07:49 PM, Eric Blake wrote: >> On 08/02/2012 11:00 AM, Eric Blake wrote: >>> On 08/02/2012 10:53 AM, Stefano Lattarini wrote: >>>> * modules/gnumakefile [Makefile.am]: The makefiles generated by >>>> Automake-NG always contain a definition of VPATH, even in non-VPATH >>>> builds (its value being simply '.' in that case). So, in the >>>> 'clean-GNUmakefile' rule, to determine whether running under a >>>> VPATH setup, compare '$(srcdir)' to '.' rather than checking whether >>>> '$(VPATH)' expands to the empty string. >>>> --- >>> >>> That fixes things with automake-ng, but what about with existing automake? >>> > The above copies an idiom mainline Automake emits in its own Makefiles > (when cleaning $(CONFIG_CLEAN_VPATH_FILES)) so I say we're safe :-) > >>>> clean-GNUmakefile: >>>> - test x'$(VPATH)' != x && rm -f $(top_builddir)/GNUmakefile || : >>>> + test '$(srcdir)' = . || rm -f $(top_builddir)/GNUmakefile >>> >>> That is, shouldn't this really be: >>> >>> case '$(srcdir)' in >>> .|'') ;; >>> *) rm -f $(top_builddir)/GNUmakefile ;; >>> esac >> >> Or am I misreading things again? I guess your complaint is that for >> in-tree builds, $(VPATH) has changed between empty and '.' depending on >> automake version, but that $(srcdir) has always been '.' for both >> settings of VPATH. >> > Exactly. Changing $(srcdir) in any way would cause major breakages > everywhere! > Ping!
There are other objections, or do I have an ACK? Thanks, Stefano
