Re: GTK_DOC_BUILD_HTML does not appear in AM_CONDITIONAL

[Simon Josefsson]

address@hidden (Ludovic Courtès) writes:

> Hi Simon,
>
> Simon Josefsson <address@hidden> writes:
>
>> Hydra Build Daemon <address@hidden> writes:
>>
>>> gtk-doc.make:53: GTK_DOC_BUILD_HTML does not appear in AM_CONDITIONAL
>>> doc/reference/Makefile.am:148:   `gtk-doc.make' included from here
>>> gtk-doc.make:58: GTK_DOC_BUILD_PDF does not appear in AM_CONDITIONAL
>>> doc/reference/Makefile.am:148:   `gtk-doc.make' included from here
>>
>> Any ideas why this happens?  There is a AM_CONDITIONAL for the symbol in
>> m4/gtk-doc.m4.  Is the build system picking up gtk-doc.m4 from the
>> system?  There is a ACLOCAL_AMFLAGS = -I m4 in the top-level
>> Makefile.am.
>
> I tracked it to the fact that Nixpkgs had GTK-Doc 1.11 whereas
> gktdoc-mkpdf is more recent.  I upgraded to 1.15, which fixed the
> problem.
>
> Besides, Nixpkgs wraps ‘aclocal’ so that it searches for m4 files in
> $ACLOCAL_PATH:
>
>   
> https://svn.nixos.org/repos/nix/nixpkgs/trunk/pkgs/development/tools/misc/automake/builder.sh
>
> However, those extra -I flags are passed *before* the user flags (see
> <https://svn.nixos.org/repos/nix/nixpkgs/trunk/pkgs/build-support/make-wrapper/make-wrapper.sh>),
> which is presumably the reason why your gtk-doc.m4 was ignored.
> Probably worth a fix.

Yes, that seems like the real source of the problem.  Even an old
GTK-Doc should not cause a build failure like that, it should just
result in a PDF-less build.

One question though: who is setting ACLOCAL_PATH?  I can't find that
variable documented in the Autoconf manual, so it isn't clear whether it
is intended to be prepended or appended to user supplied -I's.

/Simon