bug#62162: 29.0.60; xref.el doesn't compile on Emacs 26, because multi-pair setq-local

bug#62162: 29.0.60; xref.el doesn't compile on Emacs 26, because multi-pair setq-local

[João Távora]

Hi, 

In a test run of the Eglot test suite (which tests Eglot on Emacs 26.3)

   https://github.com/joaotavora/eglot/actions/runs/4405538673/jobs/7716553677#step:7:453

I'm reasonably sure that one of the failures has to do with a multi-pair
call to `setq-local` in xref.el which appeared in this commit

  commit 3573ebfa6d94411257ffefdf9eb72f508dbe502c
  Author: Juri Linkov <juri@linkov.net>
  Date:   Wed Nov 23 10:38:28 2022 +0200

That construct should not be used in xref.el because it isn't available
on Emacs 26, which xref purports to require.  In turn, this breaks Eglot
(and any other packge requiring xref) on those platforms.

João

bug#62162: 29.0.60; xref.el doesn't compile on Emacs 26, because multi-pair setq-local

[Juri Linkov]

close 62162 29.0.60
thanks

> I'm reasonably sure that one of the failures has to do with a multi-pair
> call to `setq-local` in xref.el which appeared in this commit
>
>   commit 3573ebfa6d94411257ffefdf9eb72f508dbe502c
>   Author: Juri Linkov <juri@linkov.net>
>   Date:   Wed Nov 23 10:38:28 2022 +0200
>
> That construct should not be used in xref.el because it isn't available
> on Emacs 26, which xref purports to require.  In turn, this breaks Eglot
> (and any other packge requiring xref) on those platforms.

Now this is fixed in emacs-29.  I hope Dmitry will decide
when to bump the xref package version.

bug#62162: 29.0.60; xref.el doesn't compile on Emacs 26, because multi-pair setq-local

[João Távora]

reopen 62162
thanks

Juri Linkov <juri@linkov.net> writes:

> close 62162 29.0.60
> thanks
>
>> I'm reasonably sure that one of the failures has to do with a multi-pair
>> call to `setq-local` in xref.el which appeared in this commit
>>
>>   commit 3573ebfa6d94411257ffefdf9eb72f508dbe502c
>>   Author: Juri Linkov <juri@linkov.net>
>>   Date:   Wed Nov 23 10:38:28 2022 +0200
>>
>> That construct should not be used in xref.el because it isn't available
>> on Emacs 26, which xref purports to require.  In turn, this breaks Eglot
>> (and any other packge requiring xref) on those platforms.
>
> Now this is fixed in emacs-29.  

Thanks but, I think this should have been fixed in master instead,
because that's where the :core packages are built from, AFAIK, and
emacs-29 doesn't have a problem with that construct.  It'll make no
difference soon enough, hopefully.

> I hope Dmitry will decide when to bump the xref package version.

I think you could have bumped the version, too.  Is there any reason not
to?

Closing this is premature, as the problem stands: the upcoming Eglot
1.12 will have this problem on Emacs 26.  Re-opening.  I'll bump xref.el
myself in a day or so if no-one beats me to it, and then close.

João

bug#62162: 29.0.60; xref.el doesn't compile on Emacs 26, because multi-pair setq-local

[Dmitry Gutov]

On 14/03/2023 01:38, João Távora wrote:
> reopen 62162
> thanks
> 
> Juri Linkov <juri@linkov.net> writes:
> 
>> close 62162 29.0.60
>> thanks
>>
>>> I'm reasonably sure that one of the failures has to do with a multi-pair
>>> call to `setq-local` in xref.el which appeared in this commit
>>>
>>>    commit 3573ebfa6d94411257ffefdf9eb72f508dbe502c
>>>    Author: Juri Linkov <juri@linkov.net>
>>>    Date:   Wed Nov 23 10:38:28 2022 +0200
>>>
>>> That construct should not be used in xref.el because it isn't available
>>> on Emacs 26, which xref purports to require.  In turn, this breaks Eglot
>>> (and any other packge requiring xref) on those platforms.
>>
>> Now this is fixed in emacs-29.
> 
> Thanks but, I think this should have been fixed in master instead,
> because that's where the :core packages are built from, AFAIK, and
> emacs-29 doesn't have a problem with that construct.  It'll make no
> difference soon enough, hopefully.

True enough, on both counts.

>> I hope Dmitry will decide when to bump the xref package version.
> 
> I think you could have bumped the version, too.  Is there any reason not
> to?

Extra courtesy never hurts, and after I responded (on emacs-devel) Juri 
bumped the version 40 minutes after the commit with the fix.

> Closing this is premature, as the problem stands: the upcoming Eglot
> 1.12 will have this problem on Emacs 26.  Re-opening.  I'll bump xref.el
> myself in a day or so if no-one beats me to it, and then close.

We're now waiting for the changes to be merged to master. You could do 
it yourself, if you are in a hurry.

bug#62162: 29.0.60; xref.el doesn't compile on Emacs 26, because multi-pair setq-local

[João Távora]

On Tue, Mar 14, 2023 at 12:29 AM Dmitry Gutov <dgutov@yandex.ru> wrote:

> and after I responded (on emacs-devel) Juri
> bumped the version 40 minutes after the commit with the fix.

I missed that. Great.

> > Closing this is premature, as the problem stands: the upcoming Eglot
> > 1.12 will have this problem on Emacs 26.  Re-opening.  I'll bump xref.el
> > myself in a day or so if no-one beats me to it, and then close.
>
> We're now waiting for the changes to be merged to master. You could do
> it yourself, if you are in a hurry.

I don't know how, and I'm not in that much of a hurry.  I think one day
without newest Eglot 1.12 in Emacs 26 is bearable.

Thanks,
João

bug#62162: 29.0.60; xref.el doesn't compile on Emacs 26, because multi-pair setq-local

[João Távora]

On Tue, Mar 14, 2023 at 12:43 AM João Távora <joaotavora@gmail.com> wrote:

> I don't know how, and I'm not in that much of a hurry.  I think one day
> without newest Eglot 1.12 in Emacs 26 is bearable.

I found out that it's not that hard using the admin/automerge script, which
builds and runs tests.  I did that, the merge was clean, and all tests passed,
so I pushed to master.

João

bug#62162: 29.0.60; xref.el doesn't compile on Emacs 26, because multi-pair setq-local

[João Távora]

Newest xref now in GNU ELPA.  Closing.

On Tue, Mar 14, 2023 at 6:27 PM João Távora <joaotavora@gmail.com> wrote:
>
> On Tue, Mar 14, 2023 at 12:43 AM João Távora <joaotavora@gmail.com> wrote:
>
> > I don't know how, and I'm not in that much of a hurry.  I think one day
> > without newest Eglot 1.12 in Emacs 26 is bearable.
>
> I found out that it's not that hard using the admin/automerge script, which
> builds and runs tests.  I did that, the merge was clean, and all tests passed,
> so I pushed to master.