Le 27 avr. 09 à 02:06, Ryan Schmidt a écrit :
On Apr 26, 2009, at 18:09, Rainer Müller wrote:
Rainer Müller wrote:
Thomas De Contes wrote:
---> Staging atk into destroot
[...]
/bin/sh: line 1: gtkdoc-rebase: command not found
make[2]: *** [install-data-local] Error 127
make[1]: *** [install-am] Error 2
make: *** [install-recursive] Error 1
---> Staging pango into destroot
[...]
/bin/sh: line 1: gtkdoc-rebase: command not found
make[3]: *** [install-data-local] Error 127
make[2]: *** [install-am] Error 2
make[1]: *** [install] Error 2
make: *** [install-recursive] Error 1
Could be a missing dependency on gtk-doc in both cases.
See <http://trac.macports.org/ticket/18958>.
I didn't investigate yet, but probably the fix for pango should
also be
applied to atk?
I initially thought that the current situation in e.g. glib2 was
bad: software that uses gtk-doc checks to see if gtkdoc-rebase
exists, and if so, calls it. I thought this would cause the port to
install different files, depending on whether gtk-doc was installed
or not. But it turns out that the software doesn't build the
documentation unless you use the configure arg --enable-gtk-doc.
It's off by default. Don't know what gtkdoc-rebase does, but maybe
it just rebuilds an index of available documentation, and therefore
has no real effect without --enable-gtk-doc.
The problem is that the method used to check whether gtkdoc-rebase
exists doesn't work on Mac OS X 10.4.x and earlier. The bug has been
reported to the developers of gtk-doc and will allegedly be fixed in
gtk-doc 1.12. Then each software that uses gtk-doc will have to
release a new package built with gtk-doc 1.12.
For pango, I added --enable-gtk-doc to have it rebuild the docs. The
build went rather quick so this seemed fine. I'm now investigating
glib2, which takes a *long* time to build its docs, so I don't want
to --enable-gtk-doc there. I haven't tried atk yet.
So we can either
1. declare a dependency on gtk-doc and add --enable-gtk-doc to build
or possibly rebuild the documentation,
2. declare a dependency on gtk-doc and not add --enable-gtk-doc,
just so gtkdoc-rebase exists and can be run (to no useful effect),
3. patch the Makefile(s) to use the "which" command in a way that is
compatible with Mac OS X 10.4.x and earlier, until new packages are
released built with gtk-doc 1.12 that do this for us, or
4. patch the Makefile(s) to remove the lines that check for gtkdoc-
rebase since we aready know we don't want to use it.
(1) is fine if the docs don't take long to build, (2) seems wasteful
but is easy, and (3) and (4) could be construed as being most
correct but take more time to do.
I vote for (3) or (4).
_______________________________________________
macports-users mailing list
macports-users@lists.macosforge.org
http://lists.macosforge.org/mailman/listinfo.cgi/macports-users
