bug#63674: 30.0.50; Possible miscompilation of cl-structs?

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Brian Leung]

After the latest changes to native compilation, I see a failure here: https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:7010.

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Mattias Engdegård]

> After the latest changes to native compilation, I see a failure here:
> https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:7010. 

CC:ing Andrea.

Could you help him by explaining exactly what we are looking at here? What Emacs revision is this based on, and from what revision is it a regression?

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Brian Leung]

Mattias Engdegård <mattias.engdegard@gmail.com> writes:

>> After the latest changes to native compilation, I see a failure 
>> here:
>> https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:7010. 
>
> CC:ing Andrea.
>
> Could you help him by explaining exactly what we are looking at 
> here? What Emacs revision is this based on, and from what 
> revision is it a regression?

This error occurs after Emacs commit 
9ad997cd68981fe6c6933b6977fdae23e84e6a75 (see 
https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:2657). Our 
last successful run used Emacs commit 
d4ff1d74209e97730c52ddd50c4d643c79087a33.

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Andrea Corallo]

Mattias Engdegård <mattias.engdegard@gmail.com> writes:

>> After the latest changes to native compilation, I see a failure here:
>> https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:7010. 
>
> CC:ing Andrea.

Thanks, will look at as soon as I can.

PS I merged 63679 63679 63681 as I believe are the same.

Best Regards

  Andrea

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Andrea Corallo]

Andrea Corallo <akrl@sdf.org> writes:

> Mattias Engdegård <mattias.engdegard@gmail.com> writes:
>
>>> After the latest changes to native compilation, I see a failure here:
>>> https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:7010. 
>>
>> CC:ing Andrea.
>
> Thanks, will look at as soon as I can.
>
> PS I merged 63679 63679 63681 as I believe are the same.
>
> Best Regards
>
>   Andrea

Okay I've just pushed b2b2be98da5 to fix this.

Essentially the native compiler needs `cl-typep' to be functional at
compile time for the types being defined in the compiled code.

It works for me, let's see if the fix is acceptable.

Thanks

  Andrea

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Brian Leung]

Thanks for the quick fix.

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Eli Zaretskii]

> Cc: Brian Leung <leungbk@posteo.net>, 63674@debbugs.gnu.org
> From: Andrea Corallo <akrl@sdf.org>
> Date: Wed, 24 May 2023 13:18:34 +0000
> 
> Mattias Engdegård <mattias.engdegard@gmail.com> writes:
> 
> >> After the latest changes to native compilation, I see a failure here:
> >> https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:7010. 
> >
> > CC:ing Andrea.
> 
> Thanks, will look at as soon as I can.
> 
> PS I merged 63679 63679 63681 as I believe are the same.

Your merge didn't succeed, it seems.

And people are still complaining about problems with package.el that
seem to be related to native compilation.  So could you please look
into those other issues (bug#63742, bug#63710)?

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Andrea Corallo]

Eli Zaretskii <eliz@gnu.org> writes:

>> Cc: Brian Leung <leungbk@posteo.net>, 63674@debbugs.gnu.org
>> From: Andrea Corallo <akrl@sdf.org>
>> Date: Wed, 24 May 2023 13:18:34 +0000
>> 
>> Mattias Engdegård <mattias.engdegard@gmail.com> writes:
>> 
>> >> After the latest changes to native compilation, I see a failure here:
>> >> https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:7010. 
>> >
>> > CC:ing Andrea.
>> 
>> Thanks, will look at as soon as I can.
>> 
>> PS I merged 63679 63679 63681 as I believe are the same.
>
> Your merge didn't succeed, it seems.

That's strange I think I got the message confirming that.

> And people are still complaining about problems with package.el that
> seem to be related to native compilation.  So could you please look
> into those other issues (bug#63742, bug#63710)?

Sure, I did already some work on that and today is dedicated to this
bug.

BR

  Andrea

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Andrea Corallo]

Eli Zaretskii <eliz@gnu.org> writes:

>> Cc: Brian Leung <leungbk@posteo.net>, 63674@debbugs.gnu.org
>> From: Andrea Corallo <akrl@sdf.org>
>> Date: Wed, 24 May 2023 13:18:34 +0000
>> 
>> Mattias Engdegård <mattias.engdegard@gmail.com> writes:
>> 
>> >> After the latest changes to native compilation, I see a failure here:
>> >> https://github.com/nix-community/emacs-overlay/actions/runs/5059662673/jobs/9081521991#step:6:7010. 
>> >
>> > CC:ing Andrea.
>> 
>> Thanks, will look at as soon as I can.
>> 
>> PS I merged 63679 63679 63681 as I believe are the same.
>
> Your merge didn't succeed, it seems.

Ah this is the mail I've got back form debbugs:

=====

Processing commands for control@debbugs.gnu.org:

> merge 63679 63679 63681
bug#63679: 30.0.50; package-delete: native-compile warning on startup
bug#63681: build error from native-comp: void function ("emacs-lisp/package.el" package-desc-p)
Merged 63679 63681.

>
End of message, stopping processing here.

Please contact help-debbugs@gnu.org if you need assistance.

GNU bugs database, https://debbugs.gnu.org/

=====

Not sure why it didn't work then, I assumed it did but maybe I read this
wrong.

Thanks

  Andrea

bug#63674: 30.0.50; Possible miscompilation of cl-structs?

[Eli Zaretskii]

> From: Andrea Corallo <akrl@sdf.org>
> Cc: mattias.engdegard@gmail.com,  leungbk@posteo.net,  63674@debbugs.gnu.org
> Date: Sun, 28 May 2023 10:42:15 +0000
> 
> Not sure why it didn't work then, I assumed it did but maybe I read this
> wrong.

Or maybe I was confused regarding which bugs we were talking about.

Never mind, the important thing is that now all the related bugs are
merged.