Re: glib & gtk-doc

[Owen Taylor <otaylor redhat com>]

On Thu, 2003-05-29 at 14:35, Patrick Welche wrote:
> Given the excellent documentation on line, is it really necessary
> to *require* gtk-doc to build glib from CVS? gtk-doc in turn requires
> docbook, jade, libxslt, libxml (and perl). I really don't see why
> I *have* to install docbook and jade just to build glib, would
> making this optional be considered?
> 
> As far as I can see, a first step might be to conditionally run
> gtkdocize in autogen.sh if it is found, and distributing gtk-doc.m4
> (to keep configure.in happy with a m4_include(gtk-doc.m4) so that
> --disable-gtk-doc exists) and gtk-doc.make (as it gets packaged up
> in EXTRA_DIST) -- in fact, this just "worked for me".

Nobody is forcing you to build from CVS...

The idea of requiring gtkdocize is not to force the docs
to be built but simply to get the build macros and makefile
fragments that are needed to build GTK+.

You should be able to "fake it" by only installing gtkdocize,
gtk-doc.make, and and gtk-doc.m4 from gtk-doc without the rest
the package. May be more trouble than it's worth, however.

(Question for gtk-doc-list ... maybe we should support
./configure --enable-fake-install in gtk-doc that skips
all the checks for SGML/XML and doesn't actually install the 
tools?)

> - oh - nearly worked: I had to add the patch at
> ftp://ftp.netbsd.org/pub/NetBSD/packages/pkgsrc/devel/glib2/patches/patch-aa
> (I suspect it is because NetBSD doesn't have pthread_{g,s}etschedparam
>  I don't know enough about threads to say whether those functions map to
>  pthread_attr_{g,s}etschedparam which do exist in NetBSD's pthread)
> and later gmake check failed with:

http://bugzilla.gnome.org/show_bug.cgi?id=104718

Regards,
                                       Owen