Re: ChangeLog to *VC-log*

[Stephen Berman <stephen.berman@gmx.net>]

On Wed, 9 Mar 2016 01:11:01 +0200 Dmitry Gutov <dgutov@yandex.ru> wrote:

> On 03/09/2016 01:07 AM, Stephen Berman wrote:
>
>> Ah, thanks.  I'm pretty sure it used to be the case that `C-x 4 a' would
>> automatically find the right ChangeLog file, but I suppose that changed
>> when manually maintaining ChangeLogs was abandoned.  So I guess Emacs
>> developers should now set `change-log-default-name' to top level.
>
> Probably. If you have a solid reproduction scenario, please make a bug report.
>

On Tue, 08 Mar 2016 21:01:40 -0500 Stefan Monnier <monnier@IRO.UMontreal.CA> wrote:

> AFAIK the C-x 4 a code hasn't been changed, so if there's a change in
> behavior it's most likely an accident/bug.  Please report it via M-x
> report-emacs-bug with a detailed recipe.

I found the cause of the change in behavior I observed: it's due to the
elimination of files named "ChangeLog" from the Emacs sources (there are
now only files named "ChangeLog.<n>").  In consequence, find-change-log
creates an empty file named "ChangeLog" in the immediate directory of
the file in which `C-x 4 a' was invoked, e.g. in lisp/calendar/, when
I'm editing todo-mode.el; previously, `C-x 4 a' found lisp/ChangeLog.  I
guess those who don't see this keep an non-VCS-tracked file "ChangeLog"
at top level.

If we don't return to maintaining versioned ChangeLog files, I think it
would be desirable for `C-x 4 a' to always create (or find) the file
"ChangeLog" at top level when editing Emacs sources, since this would
add the required directory levels to the commit message.  (Of course,
people could just always use `C-u C-x 4 a', but that's not as
convenient.)  Since other projects that could use `C-x 4 a' may have
different requirements on its effect, this should be conditioned by a
user option (or maybe just a plain variable), which can (and probably
should) be set in .dir-locals.el.  What do others think?

(Should I still file a bug (though it's more like a feature request), or
should we just continue the discussion here?)

Steve Berman