bug#48019: 28.0.50; Error signaled by documentation function

bug#48019: 28.0.50; Error signaled by documentation function

[Juri Linkov]

Using help commands to get docstrings from the etc/DOC file
raises an error, e.g. on typing 'C-h m':

  Debugger entered--Lisp error: (wrong-type-argument listp t)
    documentation(auto-composition-mode)
    describe-mode()
    funcall-interactively(describe-mode)
    call-interactively(describe-mode nil nil)
    command-execute(describe-mode)

This is a recent regression.

bug#48019: 28.0.50; Error signaled by documentation function

[Lars Ingebrigtsen]

Juri Linkov <juri@linkov.net> writes:

> Using help commands to get docstrings from the etc/DOC file
> raises an error, e.g. on typing 'C-h m':
>
>   Debugger entered--Lisp error: (wrong-type-argument listp t)
>     documentation(auto-composition-mode)
>     describe-mode()
>     funcall-interactively(describe-mode)
>     call-interactively(describe-mode nil nil)
>     command-execute(describe-mode)
>
> This is a recent regression.

I saw similar problems just now (not with `C-h m', but eldoc calling
`documentation' with parameters that made it bug out).  I did a "make
bootstrap", and I'm no longer seeing those problems.

If you do the same, does the problem go away for you, too?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#48019: 28.0.50; Error signaled by documentation function

[Juri Linkov]

>> Using help commands to get docstrings from the etc/DOC file
>> raises an error, e.g. on typing 'C-h m':
>>
>>   Debugger entered--Lisp error: (wrong-type-argument listp t)
>>     documentation(auto-composition-mode)
>>     describe-mode()
>>     funcall-interactively(describe-mode)
>>     call-interactively(describe-mode nil nil)
>>     command-execute(describe-mode)
>>
>> This is a recent regression.
>
> I saw similar problems just now (not with `C-h m', but eldoc calling
> `documentation' with parameters that made it bug out).  I did a "make
> bootstrap", and I'm no longer seeing those problems.
>
> If you do the same, does the problem go away for you, too?

No need to make bootstrap.  It should be enough just to restart Emacs.
So the problem is the need to restart Emacs every time after compilation.
Actually, not after every compilation, but I can't determine on what
condition this error depends.  This is a new regression appeared
in the last 2 weeks.  Before some recent change, there was no need
to restart Emacs after compilation.

bug#48019: 28.0.50; Error signaled by documentation function

[Lars Ingebrigtsen]

Juri Linkov <juri@linkov.net> writes:

> So the problem is the need to restart Emacs every time after compilation.
> Actually, not after every compilation, but I can't determine on what
> condition this error depends.  This is a new regression appeared
> in the last 2 weeks.  Before some recent change, there was no need
> to restart Emacs after compilation.

Indeed -- I just did a `C-h m' now, and now I'm also seeing the problem:

Debugger entered--Lisp error: (wrong-type-argument listp t)
  documentation(auto-compression-mode)
  describe-mode()
  funcall-interactively(describe-mode)
  call-interactively(describe-mode nil nil)
  command-execute(describe-mode)

But it doesn't seem to happen after every compilation -- perhaps this
only happens after recompiling a particular file?  Most odd.

Hm...

Could it be when compiling regenerates the etc/DOC file, perhaps?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#48019: 28.0.50; Error signaled by documentation function

[Lars Ingebrigtsen]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Could it be when compiling regenerates the etc/DOC file, perhaps?

Yup.  Running under gdb with a breakpoint I get:

#2  0x00005555557a7cbe in xsignal
    (error_symbol=XIL(0xfd50), data=XIL(0x555556f7eca3)) at lisp.h:4136
#3  0x00005555557ab6ac in xsignal2
    (error_symbol=XIL(0xfd50), arg1=XIL(0x9990), arg2=XIL(0x30)) at eval.c:1941
#4  0x000055555578bfbd in wrong_type_argument
    (predicate=XIL(0x9990), value=XIL(0x30)) at data.c:143
#5  0x00005555557b3c49 in CHECK_LIST_END (x=XIL(0x30), y=XIL(0x30))
    at lisp.h:2933
#6  0x00005555557b8a58 in Fmemq (elt=XIL(0x2aaa9be2fb48), list=XIL(0x30))
    at fns.c:1690
#7  0x0000555555796f02 in Fsnarf_documentation (filename=XIL(0x7ffff1e5e7dc))
    at doc.c:639
#8  0x0000555555796161 in reread_doc_file (file=XIL(0)) at doc.c:297
#9  0x0000555555796635 in Fdocumentation
    (function=XIL(0x2aaa9bd99738), raw=XIL(0)) at doc.c:403
#10 0x00005555557acdff in eval_sub (form=XIL(0x555556f7eeb3)) at eval.c:2520
#11 0x00005555557ac4ff in Feval (form=XIL(0x555556f7eeb3), lexical=XIL(0x30))
    at eval.c:2346
#12 0x00005555557ae7fc in funcall_subr

So...

                  if ((!NILP (Fboundp (sym))
                      || !NILP (Fmemq (sym, delayed_init)))

which is from

  Lisp_Object delayed_init =
    find_symbol_value (intern ("custom-delayed-init-variables"));

which is...  t?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#48019: 28.0.50; Error signaled by documentation function

[Lars Ingebrigtsen]

Which is this.

Stefan, this leads to errors when re-reading the etc/DOC file -- the
variable in question is assumed to always be a list.

commit 2fb271ca35ff96540c0940b0ea5281bdaac953ec
Author:     Stefan Monnier <monnier@iro.umontreal.ca>
AuthorDate: Thu Apr 22 09:32:04 2021 -0400
Commit:     Stefan Monnier <monnier@iro.umontreal.ca>
CommitDate: Thu Apr 22 09:32:04 2021 -0400

    Fix unload+reload of files using `custom-initialize-delay` (bug#47072)
    
    * lisp/custom.el (custom-initialize-delay): Don't delay if
    `custom-delayed-init-variables` has already been processed.
    * lisp/startup.el (command-line): Mark `custom-delayed-init-variables`
    as processed.

diff --git a/lisp/startup.el b/lisp/startup.el
index 4d4c65e6c4..c1267274d9 100644
--- a/lisp/startup.el
+++ b/lisp/startup.el
@@ -1175,7 +1175,7 @@ command-line
         ;; are dependencies between them.
         (nreverse custom-delayed-init-variables))
   (mapc #'custom-reevaluate-setting custom-delayed-init-variables)
-  (setq custom-delayed-init-variables nil)
+  (setq custom-delayed-init-variables t)
 
   ;; Warn for invalid user name.
   (when init-file-user



-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#48019: 28.0.50; Error signaled by documentation function

[Stefan Monnier]

> Stefan, this leads to errors when re-reading the etc/DOC file -- the
> variable in question is assumed to always be a list.

Indeed, thanks.  Should be fixed now, sorry,


        Stefan

bug#48019: 28.0.50; Error signaled by documentation function

[Lars Ingebrigtsen]

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

>> Stefan, this leads to errors when re-reading the etc/DOC file -- the
>> variable in question is assumed to always be a list.
>
> Indeed, thanks.  Should be fixed now, sorry,

Yup, works fine here not; thanks.  (So I'm closing this bug report.)

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no