bug#69631: 30.0.50; native compilation errors

bug#69631: 30.0.50; native compilation errors

[gerard.vermeulen]

Hi,

I see that native compilation of some files fails when I recompile my 
whole eln-cache.

In this case, I have in my *Async-native-compile-log* messages like:
/Users/vermeulen/opt/share/emacs/30.0.50/lisp/emacs-lisp/package.el.gz: 
Error Type or missing from typeof-types!
and the package eln file is not generated.

I see this also happen for wid-edit.el.gz and Org-mode (main repro 
checkout)
ox-texinfo, ob-core, ob-fortran, and ob-core as well as
elpa/debbugs-0.40.0.20240215.214858/debbugs-gnu.el.

GNU Emacs 30.0.50 (build 1, x86_64-apple-darwin20.6.0, NS appkit-2202.70 
Version 11.7.1 (Build 20G918)) of 2024-03-08
commit bd017175d4571e24ef1fdf84676136af1d36002d

Regards -- Gerard

bug#69631: 30.0.50; native compilation errors

[Andrea Corallo]

gerard.vermeulen@posteo.net writes:

> Hi,
>
> I see that native compilation of some files fails when I recompile my
> whole eln-cache.
>
> In this case, I have in my *Async-native-compile-log* messages like:
> /Users/vermeulen/opt/share/emacs/30.0.50/lisp/emacs-lisp/package.el.gz:
> Error Type or missing from typeof-types!
> and the package eln file is not generated.
>
> I see this also happen for wid-edit.el.gz and Org-mode (main repro
> checkout)
> ox-texinfo, ob-core, ob-fortran, and ob-core as well as
> elpa/debbugs-0.40.0.20240215.214858/debbugs-gnu.el.
>
> GNU Emacs 30.0.50 (build 1, x86_64-apple-darwin20.6.0, NS
> appkit-2202.70 Version 11.7.1 (Build 20G918)) of 2024-03-08
> commit bd017175d4571e24ef1fdf84676136af1d36002d
>
> Regards -- Gerard

Hi Gerard,

had no time yet to try master today, but I'd guess this comes from
yesterday's Stefan changes.

I'll try a whole AOT build to see if other files are impacted.

Thanks

  Andrea

bug#69631: 30.0.50; native compilation errors

[gerard.vermeulen]

On 08.03.2024 10:48, Andrea Corallo wrote:

> Hi Gerard,
> 
> had no time yet to try master today, but I'd guess this comes from
> yesterday's Stefan changes.

I think I have seen this issue for about a week, but I thought it was
a bit premature to file a bug report before the type hierarchy feature
stabilized.

> 
> I'll try a whole AOT build to see if other files are impacted.
> 
> Thanks
> 
>   Andrea

bug#69631: 30.0.50; native compilation errors

[Vincenzo Pupillo]

Hi, 
I have the same problem, but I rebuild emacs (from master, commit: bd017175d4571e24ef1fdf84676136af1d36002d) using "make boostrap". 
Last night, at home, I had no problem, but this morning:

Error: error ("../lisp/button.el" "Type or missing from typeof-types!")
  ELC+ELN  ../lisp/dos-fns.elc
  signal(error ("../lisp/button.el" "Type or missing from typeof-types!"))
  comp--native-compile("../lisp/button.el")
  batch-native-compile()
  batch-byte+native-compile()
  funcall(batch-byte+native-compile)

Thanks
Vincenzo

In data venerdì 8 marzo 2024 11:30:25 CET, gerard.vermeulen@posteo.net ha scritto:
> 
> On 08.03.2024 10:48, Andrea Corallo wrote:
> 
> > Hi Gerard,
> > 
> > had no time yet to try master today, but I'd guess this comes from
> > yesterday's Stefan changes.
> 
> I think I have seen this issue for about a week, but I thought it was
> a bit premature to file a bug report before the type hierarchy feature
> stabilized.
> 
> > 
> > I'll try a whole AOT build to see if other files are impacted.
> > 
> > Thanks
> > 
> >   Andrea
> 
> 
> 
> 

bug#69631: 30.0.50; native compilation errors

[Andrea Corallo]

Vincenzo Pupillo <v.pupillo@gmail.com> writes:

> Hi, 
> I have the same problem, but I rebuild emacs (from master, commit: bd017175d4571e24ef1fdf84676136af1d36002d) using "make boostrap". 
> Last night, at home, I had no problem, but this morning:

I confirm master is broken even for normal --with-native-compilation
(worked yesterday for me as well).

The troublesome function in button.el seems to be 'button--describe',
I'm having a look.

  Andrea

bug#69631: 30.0.50; native compilation errors

[gerard.vermeulen]

On 08.03.2024 13:46, Andrea Corallo wrote:
> 
> I confirm master is broken even for normal --with-native-compilation
> (worked yesterday for me as well).
> 
> The troublesome function in button.el seems to be 'button--describe',
> I'm having a look.
> 
When I try make bootstrap it fails with:

Loading macroexp.elc...
Wrong type argument: eieio--class, #s(built-in-class record "Abstract 
type of objects with slots." (#s(built-in-class atom "The type of 
anything but cons cells." (#s(built-in-class t "The type of everything." 
nil nil nil)) nil nil)) nil nil)
make[4]: *** [../../lisp/international/emoji-labels.el] Error 255
make[3]: *** [../lisp/international/charprop.el] Error 2
make[2]: *** [src] Error 2
***
*** "make bootstrap" failed with exit status 2.

Again, I think I have seen this for a few days (I have been grepping for 
"record").

Regards -- Gerard

bug#69631: 30.0.50; native compilation errors

[Andrea Corallo]

gerard.vermeulen@posteo.net writes:

> On 08.03.2024 13:46, Andrea Corallo wrote:
>> I confirm master is broken even for normal --with-native-compilation
>> (worked yesterday for me as well).
>> The troublesome function in button.el seems to be
>> 'button--describe',
>> I'm having a look.
>>
> When I try make bootstrap it fails with:
>
> Loading macroexp.elc...
> Wrong type argument: eieio--class, #s(built-in-class record "Abstract
> type of objects with slots." (#s(built-in-class atom "The type of
> anything but cons cells." (#s(built-in-class t "The type of
> everything." nil nil nil)) nil nil)) nil nil)
> make[4]: *** [../../lisp/international/emoji-labels.el] Error 255
> make[3]: *** [../lisp/international/charprop.el] Error 2
> make[2]: *** [src] Error 2
> ***
> *** "make bootstrap" failed with exit status 2.
>
> Again, I think I have seen this for a few days (I have been grepping
> for "record").

I'm boostrapped master many times this week with no issues, maybe we
have a different setup.

Anyway I fixed with 966d0a62a1a  the
"Type or missing from typeof-types!" issue,
this was a latent error highlited by the recent changes.

Bootstrap now progress further but fails while compiling emoji-labels.el :/

"../../src/bootstrap-emacs" -batch --no-site-file --no-site-lisp -l emoji.el -f emoji--generate-file ../../lisp/international/emoji-labels.el

Error: wrong-type-argument (eieio--class #s(built-in-class record "Abstract type of objects with slots." (#s(built-in-class atom "The type of anything but cons cells." (#s(built-in-class t "The type of everything." nil nil nil)) nil nil)) nil nil))
  signal(wrong-type-argument (eieio--class #s(built-in-class :name record :docstring "Abstract type of objects with slots." :parents (#s(built-in-class :name atom :docstring "The type of anything but cons cells." :parents (#s(built-in-class :name t :docstring "The type of everything." :parents nil :slots nil :index-table nil)) :slots nil :index-table nil)) :slots nil :index-table nil)))...

  Andrea

bug#69631: 30.0.50; native compilation errors

[Andrea Corallo]

Andrea Corallo <acorallo@gnu.org> writes:

> gerard.vermeulen@posteo.net writes:
>
>> On 08.03.2024 13:46, Andrea Corallo wrote:
>>> I confirm master is broken even for normal --with-native-compilation
>>> (worked yesterday for me as well).
>>> The troublesome function in button.el seems to be
>>> 'button--describe',
>>> I'm having a look.
>>>
>> When I try make bootstrap it fails with:
>>
>> Loading macroexp.elc...
>> Wrong type argument: eieio--class, #s(built-in-class record "Abstract
>> type of objects with slots." (#s(built-in-class atom "The type of
>> anything but cons cells." (#s(built-in-class t "The type of
>> everything." nil nil nil)) nil nil)) nil nil)
>> make[4]: *** [../../lisp/international/emoji-labels.el] Error 255
>> make[3]: *** [../lisp/international/charprop.el] Error 2
>> make[2]: *** [src] Error 2
>> ***
>> *** "make bootstrap" failed with exit status 2.
>>
>> Again, I think I have seen this for a few days (I have been grepping
>> for "record").
>
> I'm boostrapped master many times this week with no issues, maybe we
> have a different setup.
>
> Anyway I fixed with 966d0a62a1a  the
> "Type or missing from typeof-types!" issue,
> this was a latent error highlited by the recent changes.
>
> Bootstrap now progress further but fails while compiling emoji-labels.el :/
                                                 ^^^
                                                 generating

I think I'll leave this to Stefan as it looks closely related to his
recent changes and doesn't seem to involve the native compiler.

Thanks

  Andrea

bug#69631: 30.0.50; native compilation errors

[gerard.vermeulen]

On 08.03.2024 14:53, Andrea Corallo wrote:
> gerard.vermeulen@posteo.net writes:
> 
>> On 08.03.2024 13:46, Andrea Corallo wrote:
>>> I confirm master is broken even for normal --with-native-compilation
>>> (worked yesterday for me as well).
>>> The troublesome function in button.el seems to be
>>> 'button--describe',
>>> I'm having a look.
>>> 
>> When I try make bootstrap it fails with:
>> 
>> Loading macroexp.elc...
>> Wrong type argument: eieio--class, #s(built-in-class record "Abstract
>> type of objects with slots." (#s(built-in-class atom "The type of
>> anything but cons cells." (#s(built-in-class t "The type of
>> everything." nil nil nil)) nil nil)) nil nil)
>> make[4]: *** [../../lisp/international/emoji-labels.el] Error 255
>> make[3]: *** [../lisp/international/charprop.el] Error 2
>> make[2]: *** [src] Error 2
>> ***
>> *** "make bootstrap" failed with exit status 2.
>> 
>> Again, I think I have seen this for a few days (I have been grepping
>> for "record").
> 
> I'm boostrapped master many times this week with no issues, maybe we
> have a different setup.
> 
Yes.
> 
> Anyway I fixed with 966d0a62a1a  the
> "Type or missing from typeof-types!" issue,
> this was a latent error highlited by the recent changes.
> 
I confirm that you fixed the issue.
> 
> Bootstrap now progress further but fails while compiling 
> emoji-labels.el :/
> 
> "../../src/bootstrap-emacs" -batch --no-site-file --no-site-lisp -l
> emoji.el -f emoji--generate-file
> ../../lisp/international/emoji-labels.el
> 
> Error: wrong-type-argument (eieio--class #s(built-in-class record
> "Abstract type of objects with slots." (#s(built-in-class atom "The
> type of anything but cons cells." (#s(built-in-class t "The type of
> everything." nil nil nil)) nil nil)) nil nil))

My bootstrap fails almost at the same place with an error that resembles
somewhat your error above.

Thanks a lot -- Gerard

bug#69631: 30.0.50; native compilation errors

[Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors]

>> Bootstrap now progress further but fails while compiling emoji-labels.el :/
>                                                  ^^^
>                                                  generating
>
> I think I'll leave this to Stefan as it looks closely related to his
> recent changes and doesn't seem to involve the native compiler.

Hmm... I didn't bump into this problem, but I did see into another one,
so I pushed a fix, hopefully it fixed that one as well.


        Stefan

bug#69631: 30.0.50; native compilation errors

[Andrea Corallo]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>> Bootstrap now progress further but fails while compiling emoji-labels.el :/
>>                                                  ^^^
>>                                                  generating
>>
>> I think I'll leave this to Stefan as it looks closely related to his
>> recent changes and doesn't seem to involve the native compiler.
>
> Hmm... I didn't bump into this problem, but I did see into another one,
> so I pushed a fix, hopefully it fixed that one as well.

Hi,

On my system (GNU/Linux X86_64):

$ git clean -xfd && ./autogen.sh && ./configure --prefix=/home/andcor03 --with-nativecomp=aot && make -j64

I still can see the following error even if I'm on have 055e31f1d02:

"../../src/bootstrap-emacs" -batch --no-site-file --no-site-lisp -l emoji.el -f emoji--generate-file ../../lisp/international/emoji-labels.el

Error: wrong-type-argument (eieio--class #s(built-in-class record "Abstract type of objects with slots." (#s(built-in-class atom "The type of anything but cons cells." (#s(built-in-class t "The type of everything." nil nil nil)) nil nil)) nil nil))
  signal(wrong-type-argument (eieio--class #s(built-in-class :name record :docstring "Abstract type of objects with slots." :parents (#s(built-in-class :name atom :docstring "The type of anything but cons cells." :parents (#s(built-in-class :name t :docstring "The type of everything." :parents nil :slots nil :index-table nil)) :slots nil :index-table nil)) :slots nil :index-table nil)))

  Andrea

bug#69631: 30.0.50; native compilation errors

[Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors]

> $ git clean -xfd && ./autogen.sh &&
> ./configure --prefix=/home/andcor03 --with-nativecomp=aot && make -j64

With 4 cores it takes a while longer to show up.
But yes, I bumped into it after a while.
I pushed a further fix for that (which is a very similar problem).
I'm still wondering why those problems did not show up on my
earlier tests.


        Stefan

bug#69631: 30.0.50; native compilation errors

[gerard.vermeulen]

On 08.03.2024 17:21, Andrea Corallo wrote:
> Stefan Monnier <monnier@iro.umontreal.ca> writes:
> 
>>>> Bootstrap now progress further but fails while compiling 
>>>> emoji-labels.el :/
>>>                                                  ^^^
>>>                                                  generating
>>> 
>>> I think I'll leave this to Stefan as it looks closely related to his
>>> recent changes and doesn't seem to involve the native compiler.
>> 
>> Hmm... I didn't bump into this problem, but I did see into another 
>> one,
>> so I pushed a fix, hopefully it fixed that one as well.
> 
> Hi,
> 
> On my system (GNU/Linux X86_64):
> 
> $ git clean -xfd && ./autogen.sh && ./configure
> --prefix=/home/andcor03 --with-nativecomp=aot && make -j64
> 
> I still can see the following error even if I'm on have 055e31f1d02:
> 
> "../../src/bootstrap-emacs" -batch --no-site-file --no-site-lisp -l
> emoji.el -f emoji--generate-file
> ../../lisp/international/emoji-labels.el
> 
> Error: wrong-type-argument (eieio--class #s(built-in-class record
> "Abstract type of objects with slots." (#s(built-in-class atom "The
> type of anything but cons cells." (#s(built-in-class t "The type of
> everything." nil nil nil)) nil nil)) nil nil))
>   signal(wrong-type-argument (eieio--class #s(built-in-class :name
> record :docstring "Abstract type of objects with slots." :parents
> (#s(built-in-class :name atom :docstring "The type of anything but
> cons cells." :parents (#s(built-in-class :name t :docstring "The type
> of everything." :parents nil :slots nil :index-table nil)) :slots nil
> :index-table nil)) :slots nil :index-table nil)))
> 
>   Andrea

When I do make bootstrap, I get now:

Loading macroexp.elc...
Wrong type argument: eieio--class, #s(built-in-class record "Abstract 
type of objects with slots." (#s(built-in-class atom "The type of 
anything but cons cells." (#s(built-in-class t "The type of everything." 
nil nil nil)) nil nil)) nil nil)
make[4]: *** [../../lisp/international/emoji-labels.el] Error 255
make[3]: *** [../lisp/international/charprop.el] Error 2
make[2]: *** [src] Error 2
make[1]: *** [actual-bootstrap] Error 2
***
*** "make bootstrap" failed with exit status 2.

Gerard

bug#69631: 30.0.50; native compilation errors

[Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors]

> With 4 cores it takes a while longer to show up.

OK the 100% full rebuild (+ test suite) has now concluded successfully
on my end, so I believe I fixed all the problems reported.


        Stefan

bug#69631: 30.0.50; native compilation errors

[gerard.vermeulen]

On 08.03.2024 18:37, Stefan Monnier wrote:
>> With 4 cores it takes a while longer to show up.
> 
> OK the 100% full rebuild (+ test suite) has now concluded successfully
> on my end, so I believe I fixed all the problems reported.
> 
> 
>         Stefan

With commit 345cdd7a70558cd47c2ab3e124e2352debaa57cb
make bootstrap as well as a normal install show none of mine issues
reported in this thread.

So for me everything is OK, but I do not have the skills to run make 
test

Regards -- Gerard

bug#69631: 30.0.50; native compilation errors

[Andrea Corallo]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> With 4 cores it takes a while longer to show up.
>
> OK the 100% full rebuild (+ test suite) has now concluded successfully
> on my end, so I believe I fixed all the problems reported.

I confirm I've a clean bootstrap here also with
--with-native-compilation=aot

Thanks

  Andrea

bug#69631: 30.0.50; native compilation errors

[Eli Zaretskii]

> Cc: v.pupillo@gmail.com, Andrea Corallo <acorallo@gnu.org>,
>  69631@debbugs.gnu.org
> Date: Fri, 08 Mar 2024 17:54:44 +0000
> From: gerard.vermeulen@posteo.net
> 
> With commit 345cdd7a70558cd47c2ab3e124e2352debaa57cb
> make bootstrap as well as a normal install show none of mine issues
> reported in this thread.
> 
> So for me everything is OK, but I do not have the skills to run make 
> test

Thanks, I'm therefore closing this bug.