Re: ./make-dist for unicode branch

[Kenichi Handa <handa@m17n.org>]

In article <uk6cmeld0.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:

> The problem is that a tarball created with "make-dist --snapshot" out
> of a freshly checked-out CVS unicode-2 branch tree would not build
> because the files lisp/international/charprop.el and
> lisp/international/uni-*.el are generated by "make bootstrap" only if
> there's the `admin' subdirectory.  (I asked why it looks for that
> directory, but didn't see any answers.)

Ah, sorry for forgetting to answer it.  The reason is that
those files are generated from
admin/unidata/UnicodeData.txt.  This file is about 1M byte,
so I thought it was too big to be included in a tarball.
The total of generated uni-*.el is about 200K.

> Handa-san tried to fix this by checking the missing files into CVS,
> but Miles objected, saying that generated files should not be there.

> I think this is a non-issue, because make-dist is not supposed to
> produce a workable tarball unless you bootstrap first.  See
> admin/make-tarball.txt, it explicitly says to bootstrap.  As I see
> this, a snapshot should be exactly like a release tarball, i.e. it
> should include all the *.elc files etc.  So one needs to build Emacs
> before creating a snapshot.

> So I don't think these files should be in CVS, and I don't think
> there's any problem in make-dist.  Rather, before creating a snapshot,
> people should say "make bootstrap" in their sandbox.

I didn't know about admin/make-tarball.txt.  Now I agree
that we don't have to put uni-*.el in CVS tree.  Perhaps, we
should add a comment about the requirement of bootstrap at
somewhere in make-dist.

---
Kenichi Handa
handa@m17n.org