GNOME Bugzilla – Bug 661853
allow EXTRA_DIST to be predefined
Last modified: 2011-10-17 12:51:00 UTC
gtk-doc currently unconditionally assigns EXTRA_DIST, even though other parts of Makefiles may alreaady have done this. this is the case with glib, for example. We should only do an assignment if EXTRA_DIST is not already defined and do an append (+=) otherwise. Ideally, we could use ?= for this, but it seems that even though the automake documentation describes ?=, the currently implementation has a bug when ?= is mixed with +=. The ifeq route seems to work OK for both the case where EXTRA_DIST is defined before including gtk-doc's makefile and not.
> gtkdocize --help usage: gtkdocize [ --copy ] [ --docdir DIR ] [ --flavour {legacy|legacy-flat|no-tmpl|no-tmpl-flat} ] gtkdocize --flavour xxx-flat
the gtkdocize script contains this: makefile=gtk-doc.make .... case "$flavour" in legacy*) ;; no-tmpl*) makefile=gtk-doc.notmpl.make ;; *) echo "$progname: invalid value for --flavour" 1>&2 echo "$usage" 1>&2 exit 1 ;; esac and no other mentions of "flat" (except in the --help)
since the 'flat' approach is non-functioning at present (and therefore surely nobody is using it), i suggest that it is removed for favour of my patch. it seems excessive to have a separate variant for what can easier be handled with a 3-line patch.
which patch?
Created attachment 199076 [details] [review] the patch Great question. I think this was probably one of those "your bug was filed but we had difficulty attaching your patch" cases and I failed to notice it.
Thanks for the patch, that is indeed a lot better than the extra flavor. I guess i8t is gnu make only, but so are many other things anyway.
Reverted the changes and fixed it by fixing gtkdocize to handle the flavour because of automake and gnumake clashing on the conditional: https://bugzilla.gnome.org/show_bug.cgi?id=661853