Handling bugs in obsolete code (was: bug#1452: ...)

Handling bugs in obsolete code (was: bug#1452: ...)

[John Wiegley]

>>>>> Andrew Hyatt <ahyatt@gmail.com> writes:

> On Tue, Jan 5, 2016 at 10:46 PM Eli Zaretskii <eliz@gnu.org> wrote:
>> From: Andrew Hyatt <ahyatt@gmail.com>
>> 
>> Has anyone considered putting these obsolete packages in the gnu ELPA? I'm
>> not sure about the bug policy, but I'd guess that bugs shouldn't be filed
>> against ELPA packages.

>     AFAIK bugs are files against ELPA packages like they are against the
>     core Emacs. So moving to ELPA will not change this aspect of obsolete
>     packages.

>     (It also feels wrong to move them to ELPA just because they are
>     obsolete. ELPA is supposed to be home for new and advanced stuff, not
>     for obsolete stuff. If someone steps forward wanting to maintain an
>     obsolete package, then a move to ELPA might make good sense, though.)

> That's a fair point. Maybe there could be some special ELPA repository for
> obsolete packages. But what I'm mostly trying to figure out is if there is
> *any* way to get code to be completely unmaintained. We are, after all,
> trying to reduce the number of bugs (see the thread on 4k bugs) overall, and
> this is one way to do that. So the only way people would agree on right now,
> is if we remove the code entirely from emacs distribution. But I suspect
> that such a change would be rejected, even from obsolete packages, because
> someone might still be depending on them.

What if we just use an "obsolete" tag, so the bugs could be filtered out from
our running total, but they still remain open?

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2

Re: Handling bugs in obsolete code

[Andrew Hyatt]

John Wiegley <jwiegley@gmail.com> writes:

>>>>>> Andrew Hyatt <ahyatt@gmail.com> writes:
>
>> On Tue, Jan 5, 2016 at 10:46 PM Eli Zaretskii <eliz@gnu.org> wrote:
>>> From: Andrew Hyatt <ahyatt@gmail.com>
>>> 
>>> Has anyone considered putting these obsolete packages in the gnu ELPA? I'm
>>> not sure about the bug policy, but I'd guess that bugs shouldn't be filed
>>> against ELPA packages.
>
>>     AFAIK bugs are files against ELPA packages like they are against the
>>     core Emacs. So moving to ELPA will not change this aspect of obsolete
>>     packages.
>
>>     (It also feels wrong to move them to ELPA just because they are
>>     obsolete. ELPA is supposed to be home for new and advanced stuff, not
>>     for obsolete stuff. If someone steps forward wanting to maintain an
>>     obsolete package, then a move to ELPA might make good sense, though.)
>
>> That's a fair point. Maybe there could be some special ELPA repository for
>> obsolete packages. But what I'm mostly trying to figure out is if there is
>> *any* way to get code to be completely unmaintained. We are, after all,
>> trying to reduce the number of bugs (see the thread on 4k bugs) overall, and
>> this is one way to do that. So the only way people would agree on right now,
>> is if we remove the code entirely from emacs distribution. But I suspect
>> that such a change would be rejected, even from obsolete packages, because
>> someone might still be depending on them.
>
> What if we just use an "obsolete" tag, so the bugs could be filtered out from
> our running total, but they still remain open?

That would help, although it would still mean that new bugs would have
to be triaged and tagged as obsolete, as opposed to not existing at all.
If we did such a thing, it'd be nice if debbugs filtered obsolete tags
by default.

Another variant on that is to say that all bugs against obsolete packages
have "minor" severity, which would accomplish the same thing without
needing a new tag.  On the hopefully rare occasions in which the bug
really is severe (crashes emacs, corrupts data, etc) it can be have a
non-minor severity.

Re: Handling bugs in obsolete code

[John Wiegley]

>>>>> Andrew Hyatt <ahyatt@gmail.com> writes:

> That would help, although it would still mean that new bugs would have to be
> triaged and tagged as obsolete, as opposed to not existing at all. If we did
> such a thing, it'd be nice if debbugs filtered obsolete tags by default.

If minor is currently be filtered, it should be possible to filter obsolete as
well.

> Another variant on that is to say that all bugs against obsolete packages
> have "minor" severity, which would accomplish the same thing without needing
> a new tag. On the hopefully rare occasions in which the bug really is severe
> (crashes emacs, corrupts data, etc) it can be have a non-minor severity.

The advantage to having obsolete is that it would make it easier to find the
bugs we need to close whenever obsolete code is being deleted.

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2

Re: Handling bugs in obsolete code

[Glenn Morris]

I think you're over-thinking it.
The number of bugs in "obsolete" files is a tiny, insignificant
fraction, and always will be. (I'm not even sure there are any left open.)
Their influence on your stats will not be measurable.

As debbugs.gnu.org maintainer, I won't define a new global "obsolete"
tag for such a minority use. You could use a usertag, if you really want
to (or just retitle them to add eg "[obsolete]").

Re: Handling bugs in obsolete code

[CHENG Gao]

Many files were put into obsolete/ dir for long time, since 22.1 or
23.1. It does not make sense to continue keeping them there. But it's
just MPO.

How about some policy to handle this? Say keep version-1 or version-2.
For one last time, keep them all in 25.1 release tarball, and then
delete all except "Obsolete-since: 25.1/24.5", and then roll like this.

If anyone needs, they can pull from 25.1 tarball.

And maybe add instruction to each file, something like superceded-by
etc.

Re: Handling bugs in obsolete code

[Eli Zaretskii]

> From: Glenn Morris <rgm@gnu.org>
> Cc: Eli Zaretskii <eliz@gnu.org>, Andrew Hyatt <ahyatt@gmail.com>
> Date: Thu, 07 Jan 2016 02:59:32 -0500
> 
> I think you're over-thinking it.
> The number of bugs in "obsolete" files is a tiny, insignificant
> fraction, and always will be. (I'm not even sure there are any left open.)
> Their influence on your stats will not be measurable.
> 
> As debbugs.gnu.org maintainer, I won't define a new global "obsolete"
> tag for such a minority use. You could use a usertag, if you really want
> to (or just retitle them to add eg "[obsolete]").

Retitling sounds right to me, FWIW.

Re: Handling bugs in obsolete code

[John Wiegley]

>>>>> Eli Zaretskii <eliz@gnu.org> writes:

>> As debbugs.gnu.org maintainer, I won't define a new global "obsolete" tag
>> for such a minority use. You could use a usertag, if you really want to (or
>> just retitle them to add eg "[obsolete]").

> Retitling sounds right to me, FWIW.

Yes, retitling sounds good.

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2