Re: .cvsignore files

[Zefram <A.Main@dcs.warwick.ac.uk>]

>I do not think that this is really necessary.  I always build zsh in a
>separate directory so the build process does not change anything in the
>source tree with the exception of autoconf generated files.  Using
>a separate build directory is always necessary in a multiplatform
>environment.

But not all of us do this.  I almost never use a separate directory.
We must at least allow for the possibility.

>If one uses a separate directory then only config.h.in, configure and
>stamp-h.in remains.  But config.h.in, configure and the empty stamp-h.in
>are archived in the RCS files.  This is necessary since the RCS file is
>used to generate the patches between releases and not all people have
>autoconf installed on their system.  And if a file is in the RCS
>distribution it should not be put into .cvsignore.  After that no files
>remained :-).

The way I use CVS, any file that can be generated (including
config.h.in and stamp-h.in) does not go into the repository.  In fact,
I do a "make superclean" before importing sources.  I find that this is
the best way to use CVS, as one doesn't end up with redundant diffs of
generated files.

-zefram