[ODE] status of autoconf?
Jens Thiele
jens.thiele at student.uni-tuebingen.de
Thu Dec 6 06:21:02 2001
> I would include these files into CVS (and distribution
> in general):
>
> - Both configure.in *and* configure. No autoconf is
> required at installation.
> When configure.in is changed (not often), configure is
> regenerated (using autoconf),
> and both checked in.
>
> - Both Makefile.am *and* Makefile.in. No automake is
> required at installation.
> When Makefile.am is changed, Makefile.in is
> regenerated (using configure),
> and both checked in. (Note that Makefile.in contains
> no site-specific settings).
>
> - Do *not* include Makefile.
>
> - Additionally include other auxiliary files, such as:
> aclocal.m4, missing, config.guess, config.sub,
> install-sh, autogen, etc.
>
>
Well I think that is really a matter of taste. A distribution tarball should
of course include all the files => no need for autoconf/automake to compile
but in cvs it is really a matter of taste. If you assume the people getting
the code from cvs have autoconf/... - i would leave it out (you just
have to type
./autogen) - otherwise I would perhaps include it.
> Yes, this sounds like a good plan. If tests are
> created which work even without
> graphics, they should be put in a new directory, and
> compiled even
> with --enable-drawstuff=no.
> It is trickier if you want such a test to be built
> without drawstuff,
> but *with* drawstuff *if* available. I see this
> possible if code using drawstuff
> are isolated using #ifdefs. (The availablilty of
> drawstuff (ODE_DS)
> should be included in the generated config.h,
> currently it isn't).
>
yep - how should the define be called ? (perhaps not ODE_DS ;-)
--
email karme@unforgettable.com
IRC Karme
HP http://www-ti.informatik.uni-tuebingen.de/~thiele