GNU bug report logs - #67900
30.0.50; Emacs Crahes When Executing Command `consult-buffer'

Previous Next

Full log

View this message in rfc822 format

From: Andrea Corallo <acorallo <at> gnu.org>
To: Chang Xiaoduan <drcxd <at> sina.com>
Cc: 67900 <at> debbugs.gnu.org, Eli Zaretskii <eliz <at> gnu.org>
Subject: bug#67900: 30.0.50; Emacs Crahes When Executing Command `consult-buffer'
Date: Tue, 09 Jan 2024 04:58:00 -0500

Chang Xiaoduan <drcxd <at> sina.com> writes:

> Hello Andrea,
>
> Andrea Corallo <acorallo <at> gnu.org> writes:
>
>> You mean with the whole compilation unit is compiled at speed 2?
>
> No, all the `defun', `cl-defun', and `defmacro' are declared with
> `(speed 1)`.
>
>> Mmmh not sure I understand, if you macro expanded all the instances of
>> this macro how can the macro matter given it's never called?
>
> I am very curious about this point as well. At first, I expand the macro
> without the declare form in the function definition in the macro
> definition. Then I add the declare form to the functions generated by
> the macro expansion. To my surprise, the crash still occurs.
>
> In another test, I add the declare form to the function definition in
> the macro definition, and expanded the macros. The crash can not be
> reproduced. I remove all the declare form in the generated functions,
> and the crash still can not be reproduced.
>
> Thus I come to the conclusion that the declare form in the function
> definition in the macro definition is what triggers the crash.
>
> I do not know Emacs lisp much so I convinced myself that maybe macros in
> Emacs lisp is different from macros in C++ and even if there is no
> instance of using the macro it still get compiled in the native
> compilation. Thus, I just reported what I have observed to you.
>
>>
>> I think we have to really understand what's happening here and if what
>> you've observed is reproducible cause it does not make much sense to me
>> ATM.
>>
>> Thanks for your investigation
>>
>>   Andrea
>
> I guess you do not use Windows?

I don't, not usually at least.

> Maybe I can setup a Windows VM so that I
> can sent the VM to you and you can see the crash by yourselves.

Yes I could have a look.

> I also
> wonder is there any way that we can check the C source code generated
> from the Emacs lisp code before it is sent to the C compiler?

Yep, see `native-comp-debug'.

Thanks

  Andrea

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.