auctex-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[AUCTeX-devel] Re: [Bug-AUCTeX] 2005-06-09; `make' tries to run texi2dv


From: David Kastrup
Subject: [AUCTeX-devel] Re: [Bug-AUCTeX] 2005-06-09; `make' tries to run texi2dvi now on Windows
Date: Mon, 13 Jun 2005 15:45:29 +0200
User-agent: Gnus/5.11 (Gnus v5.11) Emacs/22.0.50 (gnu/linux)

David Kastrup <address@hidden> writes:

> Christian Schlauer <address@hidden> writes:
>
>> Remember to cover the basics, that is, what you expected to happen and
>> what in fact did happen.
>>
>
> This means that the Makefile fires in order to generate auctex.dvi
> from auctex.texi.  I don't think that the normal run of "make"
> should produce a dvi file.  We don't install dvi files as part of
> the tree, and so we should not create it.  If we chose to have it as
> part of the regular installation, then it should already be present
> in the tarball premade, like auctex.info is.  So this would appear
> to be a bug in the targets.
>
> I'll take a look at how to fix this.

Looks like we have a pretty solid mess here, and I would like opinions
of other developers how to clear this up.

Here is my take on the matter:

I would like to remove all traces of CVS from the tarball.  That means
that I don't want to distribute README.CVS and autogen.sh (and
.cvsignore files).  At the same time, I don't want to deliver a
tarball that is not a self-sufficient source in itself.

That means that I want everything that is done by autogen.sh to be
replicatable by the Makefiles.  Those get generated from Makefile.in
by configure, and we do put a compiled copy of configure into the
tarballs, so we are ready to go here.

Of course, we don't have a compiled copy of configure in the CVS, so
we still need autogen.sh to bootstrap from there.

I think that the effects of autogen.sh should be pretty much the
effects of "make dist", which they now definitely are not.  And "make
dist", I think, should not need to do more than to run autoconf (with
cache directory cleanup) in top and preview directory, descend into
the doc directory and do a "make dist" there.  It should not produce
anything we don't place in the tarball.

Before actually packaging the tarball from an exported archive, we
should remove the files

.cvsignore
doc/.cvsignore
preview/.cvsignore
preview/latex/.cvsignore
autogen.sh
README.CVS
preview/autogen.sh

Actually, I think we can fold preview/autogen.sh into the top
autogen.sh right now, anyway.  There is no need to bootstrap an
AUCTeX-less tree as far as I can see, and it does not work, to boot.

I think that this list of files would be more or less hardcoded into
the tarball target (through some fixed Makefile variable) which also
does the export from CVS.  I don't think that either make dist or make
distclean should care about those files.  Since the Makefile doing the
deletions is _not_ the same Makefile as that in the tree checked out
for the tarball (which does not even exist...), it does not seem to
make sense to encode this list into a maintainer-clean or similar
target.  The cleanup is necessary in a different tree from that where
the Makefile resides.

Do people feel comfortable with the gist of the proposal?

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum




reply via email to

[Prev in Thread] Current Thread [Next in Thread]