Re: [Emacs-diffs] Changes to emacs/lib-src/makefile.w32-in,v

Re: [Emacs-diffs] Changes to emacs/lib-src/makefile.w32-in,v

[Juanma Barranquero]

On 11/27/06, Eli Zaretskii <eliz@gnu.org> wrote:

> +# emacsclient.$(O) depends on makefile.w32-in because makefile.w32-in
> +# can be edited to define VERSION string, which is part of ECLIENT_CFLAGS.

I didn't think of it because makefile.w32-in's VERSION number is only
modified at release time. As it was, it would show the wrong version
number only for developers compiling (not bootstrapping) from the CVS,
and only when emacsclient.c is not modified in a release. But you're
right in changing it, of course.

                    /L/e/k/t/u

Re: [Emacs-diffs] Changes to emacs/lib-src/makefile.w32-in,v

[Eli Zaretskii]

> Date: Mon, 27 Nov 2006 10:42:00 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: "Emacs Devel" <emacs-devel@gnu.org>
> 
> On 11/27/06, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > +# emacsclient.$(O) depends on makefile.w32-in because makefile.w32-in
> > +# can be edited to define VERSION string, which is part of ECLIENT_CFLAGS.
> 
> I didn't think of it because makefile.w32-in's VERSION number is only
> modified at release time. As it was, it would show the wrong version
> number only for developers compiling (not bootstrapping) from the CVS,
> and only when emacsclient.c is not modified in a release. But you're
> right in changing it, of course.

It's not that I'm so clever, it's just that after "cvs up" and
"configure", saying "make" didn't rebuild anything.

Re: [Emacs-diffs] Changes to emacs/lib-src/makefile.w32-in,v

[Juanma Barranquero]

On 11/27/06, Eli Zaretskii <eliz@gnu.org> wrote:

> It's not that I'm so clever, it's just that after "cvs up" and
> "configure", saying "make" didn't rebuild anything.

:)

                    /L/e/k/t/u

Re: [Emacs-diffs] Changes to emacs/lib-src/makefile.w32-in,v

[Juanma Barranquero]

On 11/27/06, Juanma Barranquero <lekktu@gmail.com> wrote:

> > +# emacsclient.$(O) depends on makefile.w32-in because makefile.w32-in
> > +# can be edited to define VERSION string, which is part of ECLIENT_CFLAGS.

BTW, shouldn't it depend on lib-src/makefile instead of
lib-src/makefile.w32-in? If the user does not run configure, makefile
will have the old version anyway.

                    /L/e/k/t/u

Re: [Emacs-diffs] Changes to emacs/lib-src/makefile.w32-in,v

[Eli Zaretskii]

> Date: Mon, 27 Nov 2006 16:49:13 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: "Emacs Devel" <emacs-devel@gnu.org>
> 
> BTW, shouldn't it depend on lib-src/makefile instead of
> lib-src/makefile.w32-in?

That'd be a nuisance, I think: people run configure even if
makefile.w32-in didn't change (well, I do), just to be sure.

> If the user does not run configure, makefile will have the old
> version anyway.

What you are actually saying is that makefile in each directory should
depend on the respective makefile.w32-in, and if the latter was
changed, nt/configure.bat should run automatically.  You are probably
right, but let's not do this before the release.

(While at that, we should also avoid recreating makefile's and other
files produced by nt/configure.bat if the new version would be
identical to the old one, to avoid unnecessary recompilation.)

Re: [Emacs-diffs] Changes to emacs/lib-src/makefile.w32-in,v

[Juanma Barranquero]

On 11/27/06, Eli Zaretskii <eliz@gnu.org> wrote:

> That'd be a nuisance, I think: people run configure even if
> makefile.w32-in didn't change (well, I do), just to be sure.

The only (tiny) problem with depending on makefile.w32-in, is that if
the user forgets to run configure, he will see emacsclient being
compiled, but the version number will still be old. It's a bit
misleading.

Just a nitpick, anyway. The main case to consider is whether users
(not developers building from the CVS tree) will get the right number
after bootstrapping the first time, and they will.

> What you are actually saying is that makefile in each directory should
> depend on the respective makefile.w32-in, and if the latter was
> changed, nt/configure.bat should run automatically.

A rule in makefile making it depend on makefile.w32-in? Nice idea :)

> You are probably
> right, but let's not do this before the release.

No, of course not!

> (While at that, we should also avoid recreating makefile's and other
> files produced by nt/configure.bat if the new version would be
> identical to the old one, to avoid unnecessary recompilation.)

That shouldn't be too tricky, if you can count on a reliable way to
compare the old and new files. (Do we require diff right now? I don't
think so.)

                    /L/e/k/t/u

Re: [Emacs-diffs] Changes to emacs/lib-src/makefile.w32-in,v

[Eli Zaretskii]

> Date: Mon, 27 Nov 2006 17:20:55 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: emacs-devel@gnu.org
> 
> > (While at that, we should also avoid recreating makefile's and other
> > files produced by nt/configure.bat if the new version would be
> > identical to the old one, to avoid unnecessary recompilation.)
> 
> That shouldn't be too tricky, if you can count on a reliable way to
> compare the old and new files. (Do we require diff right now? I don't
> think so.)

We could write a small program that does what the move-if-changed
script does, and use that.