Re: 16-ac-lang-preproc-require.patch

[Pavel Roskin]

Hello, Akim!

> | I'm not sure that we really want to go ahead after that warning. Howvalid
> | "configure" will be if we ignore it?
> 
> Maybe I didn't think enough about this issue, but since I have no
> experience of Fortran...
> 
> BTW, this weekend I had a stupid idea: how about introducing
> `Preprocessed Fortran 77' as a language, which has ac_ext=.F?

Please don't do it. I would prefer to accept the ideas only from the
people who have any real experience in this area.

By the way, I believe that Autoconf/Automake/Libtool should move in the
direction of creating wrappers around tools. The wrappers for every tool
could implement many functions, including dependency tracking, option
conversion, workarounds for bugs, deansification and (for Fortran)
providing preprocessor for the compilers that don't have it.

But it's all the distant future. For now I would stick only with highly
demanded changes.

> | Well, we don't test the macros that are AC_REQUIRE'd. This doesn't seem to
> | be a very clean appoach, since the caller can cleanup for the callee.
> | mktests.sh may need to be changed.
> 
> I understand your concern, but I'm also afraid to see the test suite
> slowing down too much.  Personally I prefer it as is currently.

It's a valid concern. I'll check those macros manually. Also we could 
consider something like "make check SLOWTEST=yes"

Regards,
Pavel Roskin