bug#67683: 30.0.50; [PATCH] Fix 'M-.' on Emacs C sources

[Manuel Giraud]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Manuel Giraud <manuel@ledu-giraud.fr>
>> Cc: 67683@debbugs.gnu.org
>> Date: Thu, 18 Jan 2024 22:10:48 +0100
>> 
>> Eli Zaretskii <eliz@gnu.org> writes:
>> 
>> [...]
>> 
>> >> I'm not using Edebug but since I currently get "(void-function nil)"
>> >> error I did a 'toggle-debug-on-error' and this error triggers when
>> >> calling 'tags-included-tables' from 'tags-table-extend-computed-list'
>> >> 
>> >> What is really weird is that if I set myself into the TAGS buffer and
>> >> evaluate (tags-included-tables), it works and returns a list of
>> >> filenames.
>> >
>> > And if you type "C-u M-." in the TAGS buffer, then you can type a
>> > symbol, and Xref shows its definition allright?  IOW, does M-. work
>> > from the TAGS buffer?
>> 
>> No "C-u M-." does not work from the TAGS buffer with the same error.
>
> So then the conclusion is that something in the way
> tags-included-tables is invoked from tags-table-extend-computed-list
> (or its callers in the M-. scenario) breaks tags-included-tables, even
> though invoking tags-included-tables directly works as intended,
> right?  If so, I suggest to look up the call-stack chain and try to
> find what is it that breaks tags-included-tables.

Thanks.  I have found the problem even though I don't what causes it.
It came from the "lisp/TAGS" buffer: in this one, evaluating
(tags-included-tables) returns an error.  I just had to close this
buffer and now xref is working as expected.

But I really don't know what caused this buffer to be broken in this
regard.  Anyway thanks (again) for your kind help.
-- 
Manuel Giraud