bug#25590: Remove build number from emacs-version variable

[Noam Postavsky]

On Mon, Feb 6, 2017 at 10:32 AM, Eli Zaretskii <eliz@gnu.org> wrote:
>> From: Glenn Morris <rgm@gnu.org>
>> Cc: 25590@debbugs.gnu.org
>> Date: Sun, 05 Feb 2017 18:45:34 -0500
>>
>> Eli Zaretskii wrote:
>>
>> > If we do the above, we should make sure version-comparison functions
>> > treat something like "25.2 (build 4)" correctly.
>>
>> I disagree that this is desirable.
>
> Can you explain why?  My reasoning was that if "25.2.4" is replaced by
> "25.2 (build 4)", then Lisp code that compares version (e.g., I have
> such code in my .emacs) will not work properly unless the
> version-comparison functions are updated to support such values.

I think the proposal is to go from this:

emacs-version ;=> "24.5.1"
(emacs-version) ;=> "GNU Emacs 24.5.1 (i686-pc-mingw32)
 of 2015-04-11 on LEG570"

to this

emacs-version ;=> "24.5"
emacs-build-number ;=> 1
(emacs-version) ;=> "GNU Emacs 24.5 (Build 1) (i686-pc-mingw32)
 of 2015-04-11 on LEG570"

i.e., there are no changes in version-comparibility:
 (version< emacs-version "24.5") continues to work,
 (version< (emacs-version) "24.5") continues to fail.

>
>> > It's actually part of a version number (which many other projects
>> > have, e.g., GDB just released version 7.12.1), except that we never
>> > release such versions, they exist only on end-users' machines.  Other
>> > than that, there's nothing in it which is special to Emacs.
>>
>> I don't understand the comparison. A micro version number
>> (major.minor.micro) is not the same thing as Emacs's build number.
>
> I think it is.
>

GDB 7.12.1 is a public release, the official NEWS file documents the
differences between that and 7.12. For Emacs, there is no well-defined
difference between 25.1.1 and 25.1.2, it could be different for every
person who builds it.