Re: autoconf and cl.exe

[Braden McDaniel]

"Paul Eggert" <address@hidden> wrote in message
news:address@hidden
> "Braden McDaniel" <address@hidden> writes:
>
> > The Coin <http://coin3d.org> folks have solved this by using a wrapper
that
> > translates the POSIX arguments to cl-ese. I'm wondering (1) if such a
thing
> > as this would be considered a useful addition to autoconf itself;
>
> Yes.  However, we'd need help to maintain it.  I don't use cl or
> Windows, for example.
>
> > and (2), if so, what kind of requirements it would need to meet
> > (i.e., presumably it ought to be written in sh).
>
> That would normally be one requirement.  However, if cl in practice is
> always run on a platform that uses some other shell, it'd be OK to use
> that shell.  I.e., if everybody who uses cl uses command.exe (or
> whatever the other shell is), and if sh is not universally available,
> then you might as well use command.exe.

If sh is not available, presumably there would be substantial difficulty in
using autoconf/configure at all.

In practice, I think there are very few projects using cl with autoconf. But
I want to use autotools, and building with cl is a requirement for the
project I'm working on. So hopefully I can come up with a general solution
to the problem.

> A couple of other requirements:
>
> * The changes should be decoupled from the rest of Autoconf.  We
>   shouldn't have to have the equivalent of "#ifdef DOS" all over the
>   place.

Of course; avoiding that kind of thing is very much the point of the
wrapper.

> * The author of the changes would need to sign the copyright over to
>   the FSF.

I don't anticipate a problem.

I'm glad there's interest.

--
Braden McDaniel <address@hidden>
Software Engineer, Object Sciences Corporation