Re: EMACS Ada Mode Maintenance

[john@assen.demon.co.uk.nospam (John McCabe)]

dewar@gnat.com wrote:

>In article <36b9f769.3385631@news.demon.co.uk>,
>  john@assen.demon.co.uk.nospam (John McCabe) wrote:
>> dewar@gnat.com wrote:
>> Although I would stand by such a statement, mainly with
>> respect to new features, there is a subtlety in this
>> quote in that it doesn't state anywhere that general
>> features of currently available GNU software should be
>> removed and replaced with product specific code (I refer
>> in particular of course to the use of gnatstub in
>> ada-make-body).

>If you want to know Richard Stallman's position here, it is
>that any change to Ada mode should be evaluated ONLY with
>respect to its effect on the users of GNU software.

That's perfectly reasonable.

The problem you have really is that if you make ada-mode awkward to
use with other compilers, then people are simply going to think, "well
is it worth the effort?" and move away from GNU software altogether,
especially as many compilers (especially on Windows NT systems) come
with well integrated GUIs including editors - the editor may be pretty
poor (like Aonix ObjectAda's), but it is the integration with the rest
of they system that can be important. Please note I am not suggesting
that you are likely to do this deliberately of course.

>He objects to any effort or consideration being given to users of
>non-free software, so if you ask for a policy decision here from
>the "official" source, you will in fact get a rather definite
>view along these lines.

>In this particular case, the issue is entirely whether the stub facility
>is improved with respect to its use with GNAT.

>This does not stop someone developing other code that uses some
>other approach usable with another compiler, and if this is also
>useful with GNAT, does not impede its inclusion in the official Ada mode.

>In this particular case, it seemed clear that the change to using
>gnatstub improves the behavior of this feature in the GNU system,
>and that therefore the change is beneficial to users of the GNU system.

I haven't used gnatstub, so I cannot comment, but ada-make-body did
have a lot of problems in V2.27 (that your Version 3.1 is based on)
although I believe that was one of the areas Rolf worked on for V2.29.
I can only assume that gnatstub does a better job, but not everyone
who has Emacs will have gnatstub, and the ada-make-body code was
reasonably serviceable in many situations. The thing to remember is
that ada-make-body is called through a hook into find-file, and that
it would have been a simple matter just to leave ada-make-body as it
was, and change the hook to point to a function that used gnatstub.

>This is, always has been, and will continue to be, the primary
>criterion by which changes to GNU software are judged by the FSF.
>You sale that Rolf was reluctant to make changes that were GNAT
>specific. In fact I suspect that if Richard Stallman had know this,
>he might not have been very happy with the situation.

This was an isolated case regarding ada-xref, I cannot comment on his
general view towards GNAT.

>But as I said in my last note, in practice, overall, I think everyone,
>even those not using GNAT, will find that having someone able to work
>on Ada mode who is fully supported rather than doing it on a volunteer
>basis will work out very well.

I believe it will also. 

>I suggest everyone watch and see how this works over the next few
>months, and withold judgments till we have that experience.

I will certainly be watching.

Although it may sound that way, I'm not trying to be negative about
all this. I think it's probably going to be good for ada-mode, however
in my view, if I have to considerably hack ada-mode.el to make it work
with a compiler I'm using, rather than just replace a hook or function
call with one based on what is already available for GNAT, then I may
decide that the effort involved is not worth it.

I can't see that happening really, but I think it needs to be
considered. GNU software after all needs its users for support, and if
users move away from it on the grounds that it is irreconcilably tied
in to other GNU software, then it is bad news.


Best Regards
John McCabe <john@assen.demon.co.uk>