Re: [Libreoffice-commits] .: configure.ac

On Mon, Jan 21, 2013 at 5:55 AM, Rene Engelhard <rene@debian.org> wrote:
> Can we please stop this "auto-deect-everything-and-disable-stuff-silently"
> nonsense?

Well it is 'auto-select-and-enable-stuff-based-on-what-is-avaialble'
nonsense :-)

> Imagine the following scenario.
>
> Default build without doxygen but *with* the SDK. (e.g. binary-only builds).
> The Debian buildds won't install doxygen because it's only needed for
> arch-indep stuff (and the arch-dep part uses -without-doxygen). So  the
> binary-only build will see no doxygen -> builds no SDK -> bad.

distro-config build should be explicit and not rely on implicit
'default' values, since these will change eventually over time.

> There *IS* already -without-doxygen for people who want the SDK but
> not doxygen.

There *IS* --enable-odk for people who want the SDK and it *IS*
honored if specified.

The only case that this patch impact is:

Before: if you did _not_ specified --enable-odk _and_ you did not
specify --with-dogygen=/path or --without-doxygen
then the configure would imply --enable-odk=yes and fail if doxygen is
not in the PATH

After: In the same circumstance the configure default --enable-odk to no

iow: before: if you did plain ./autogen.sh and did not have doxygen =>
enable-odk=yes and error because doxygen is not in the PATH
       after: if you do plain .autogen.sh if doxygen is not in the
PATH enable-odk is defaulted to no. => no error

All other behavior of the combinations of enable-odk /with-doxygen are
left unchanged....

> Oh, and it wozld make more sense to get such changes reviewed by Linux
> people or distro packjagers who heavily rely on configure switches in
> their packaging...

I purposefully pushed that patch to gerrit and did not push it
directly. what else can I do to get review-before-hitting-the-tree ?

Norbert