bug#62248: 29.0.60; nil obsolete

bug#62248: 29.0.60; nil obsolete

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

Type:

  C-h v nil RET

Emacs will describe `nil' as obsolete since 29.1.
That is absurd.

bug#62248: 29.0.60; nil obsolete

[Eli Zaretskii]

> Date: Sat, 18 Mar 2023 10:48:07 +0800
> From:  Po Lu via "Bug reports for GNU Emacs,
>  the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>
> 
> Type:
> 
>   C-h v nil RET
> 
> Emacs will describe `nil' as obsolete since 29.1.

I cannot reproduce this.  I tried MS-Windows and GNU/Linux.  I get:

  nil’s value is nil

  Not documented as a variable.

    Probably introduced at or before Emacs version 1.6.

which looks reasonable enough.

bug#62248: 29.0.60; nil obsolete

[Michael Welsh Duggan]

Po Lu via "Bug reports for GNU Emacs, the Swiss army knife of text
editors" <bug-gnu-emacs@gnu.org> writes:

> Type:
>
>   C-h v nil RET
>
> Emacs will describe `nil' as obsolete since 29.1.
> That is absurd.

Happens for me too, but not with -Q.  So some common package is causing
this, likely by accident.

-- 
Michael Welsh Duggan
(md5i@md5i.com)

bug#62248: 29.0.60; nil obsolete

[Eli Zaretskii]

> Cc: luangruo@yahoo.com
> From: Michael Welsh Duggan <mwd@md5i.com>
> Date: Sat, 18 Mar 2023 02:52:57 -0400
> 
> Po Lu via "Bug reports for GNU Emacs, the Swiss army knife of text
> editors" <bug-gnu-emacs@gnu.org> writes:
> 
> > Type:
> >
> >   C-h v nil RET
> >
> > Emacs will describe `nil' as obsolete since 29.1.
> > That is absurd.
> 
> Happens for me too, but not with -Q.  So some common package is causing
> this, likely by accident.

Can you step through help-fns--var-obsolete and tell what you see
there?  E.g., does (get nil 'byte-obsolete-variable) return non-nil?

If so, how about running Emacs with a breakpoint in Fput and condition
that symbol == Qnil (assuming your Emacs is not configured to
represent Lisp objects as structs)?

bug#62248: 29.0.60; nil obsolete

[Michael Welsh Duggan]

Eli Zaretskii <eliz@gnu.org> writes:

>> Cc: luangruo@yahoo.com
>> From: Michael Welsh Duggan <mwd@md5i.com>
>> Date: Sat, 18 Mar 2023 02:52:57 -0400
>> 
>> Po Lu via "Bug reports for GNU Emacs, the Swiss army knife of text
>> editors" <bug-gnu-emacs@gnu.org> writes:
>> 
>> > Type:
>> >
>> >   C-h v nil RET
>> >
>> > Emacs will describe `nil' as obsolete since 29.1.
>> > That is absurd.
>> 
>> Happens for me too, but not with -Q.  So some common package is causing
>> this, likely by accident.
>
> Can you step through help-fns--var-obsolete and tell what you see
> there?  E.g., does (get nil 'byte-obsolete-variable) return non-nil?
>
> If so, how about running Emacs with a breakpoint in Fput and condition
> that symbol == Qnil (assuming your Emacs is not configured to
> represent Lisp objects as structs)?

I traced it, in my instance, to this line from url-gw.el:

(make-obsolete-variable url-gw-rlogin-obsolete-warned-once nil "29.1")

Maybe `make-obsolete-variable' should warn or error with nil as its
initial argument.

-- 
Michael Welsh Duggan
(md5i@md5i.com)

bug#62248: 29.0.60; nil obsolete

[Eli Zaretskii]

> Cc: luangruo@yahoo.com, 62248@debbugs.gnu.org
> Date: Sat, 18 Mar 2023 09:05:11 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> 
> If so, how about running Emacs with a breakpoint in Fput and condition
> that symbol == Qnil (assuming your Emacs is not configured to
> represent Lisp objects as structs)?

Or how about adding

  if (NILP (symbol) || EQ (symbol, Qt))
    emacs_abort ();

to Fput, and then run with this under GDB?

bug#62248: 29.0.60; nil obsolete

[Eli Zaretskii]

> From: Michael Welsh Duggan <mwd@md5i.com>
> Cc: Michael Welsh Duggan <mwd@md5i.com>,  62248@debbugs.gnu.org,
>   luangruo@yahoo.com
> Date: Sat, 18 Mar 2023 03:15:11 -0400
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > If so, how about running Emacs with a breakpoint in Fput and condition
> > that symbol == Qnil (assuming your Emacs is not configured to
> > represent Lisp objects as structs)?
> 
> I traced it, in my instance, to this line from url-gw.el:
> 
> (make-obsolete-variable url-gw-rlogin-obsolete-warned-once nil "29.1")

OK, I already found (by different means) that one, and fixed it (and
another similar one in url-misc).

> Maybe `make-obsolete-variable' should warn or error with nil as its
> initial argument.

We could install something like that on master, yes.  But then we
should also warn if the symbol is t, I think.

bug#62248: 29.0.60; nil obsolete

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

Eli Zaretskii <eliz@gnu.org> writes:

> Can you step through help-fns--var-obsolete and tell what you see
> there?  E.g., does (get nil 'byte-obsolete-variable) return non-nil?

Yes, (get nil 'byte-obsolete-variable) returns:

  (nil nil "29.1")

> If so, how about running Emacs with a breakpoint in Fput and condition
> that symbol == Qnil (assuming your Emacs is not configured to
> represent Lisp objects as structs)?

I will try that.

bug#62248: 29.0.60; nil obsolete

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

Michael Welsh Duggan <mwd@md5i.com> writes:

> I traced it, in my instance, to this line from url-gw.el:
>
> (make-obsolete-variable url-gw-rlogin-obsolete-warned-once nil "29.1")

Ah, thanks!  This saves me quite some trouble.
This should be:

  (make-obsolete-variable 'url-gw-rlogin-obsolete-warned-once nil "29.1")

bug#62248: 29.0.60; nil obsolete

[Stefan Kangas]

Version: 30.1

Eli Zaretskii <eliz@gnu.org> writes:

>> Maybe `make-obsolete-variable' should warn or error with nil as its
>> initial argument.
>
> We could install something like that on master, yes.  But then we
> should also warn if the symbol is t, I think.

I made it error in that case, so I'm closing the bug.