On 25 Aug 2001 11:35:59 +0200, Christoph Egger wrote:
>
> Hi!
>
> I have tested "make distcheck" on libggimisc, libgii and libggi.
> Worked fine for me.
>
> But I have also realized that, every target, where configure says "no"
> doesn't come into the source tarball.
Yep. I just realized this myself. Irritating. I've been puzzling over
how this is supposed to work - I'm sure that GGI isn't the only project
that works this way.
> What I want to say is, that these source-tarballs (created with make
> distcheck) are very platform specific.
>
> Example: On my Linux machine, configure says no to directx, vgl and some
> other targets, requiring non-installed packages (i.e. aalib).
>
> So, when I do a make distcheck and packagers uses the resulting source
> tarball, then FreeBSD- and DirectX- packagers are pissed off, because
> there is no directx-code and no vgl-code, which they need. Further,
> users, that uses aalib as a regular base, are also pissed off, because
> there is no aalib-target.
>
> My libgii source tarball would piss off joystick users, for example.
>
>
> A FreeBSD -user, that creates a source tarball with "make distcheck" will
> only contain everything he needs, that it fits on freebsd systems.
>
>
> In short words: We have to create a source tarball for each platform, GGI
> is ported to, to be able to make a _working_ release for that platform.
>
>
> So, I suggest, making a simple tarball containing the _whole_ source tree
> (don't forget to call "autogen.sh" to not miss the configure script) for
> all plattforms should be enough. Of course, the tarball has the name
> convention <libname>-<major>.<minor>.<micro>.tar.<compression>.
I'm trying to come up with a solution that is a hybrid of "make
distcheck" and including all the source. This is one reason why I asked
about the "EXTRA_SUBDIRS" variable that I noticed and found unused.
> Our packagers should do the "make distcheck" then.
>
> Comments?
I'm hoping that the magic fairy of the north will hand me a solution so
that my brain doesn't have to think of some contrived method that might
confuse others later on. If past history is any indication then I'm not
optimistic that this might happen.
>
> CU,
>
> Christoph Egger
> E-Mail: [EMAIL PROTECTED]
>
--
Thayne Harbaugh
