Re: master 977f3f27c5 2/2: Don't issue warnings for Unicode quotes for now

Re: master 977f3f27c5 2/2: Don't issue warnings for Unicode quotes for now

[Stefan Kangas]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> branch: master
> commit 977f3f27c549db27a2d6fb33e0112b03f9b57371
> Author: Lars Ingebrigtsen <larsi@gnus.org>
> Commit: Lars Ingebrigtsen <larsi@gnus.org>
>
>     Don't issue warnings for Unicode quotes for now
>
>     * lisp/emacs-lisp/bytecomp.el (byte-compile-docstring-style-warn):
>     Remove warning for "Unicode quotes" for now (bug#55780).

How about keeping the warning but making it optional instead?  That
way, we could still use it in emacs.git.

Re: master 977f3f27c5 2/2: Don't issue warnings for Unicode quotes for now

[Lars Ingebrigtsen]

Stefan Kangas <stefan@marxist.se> writes:

> How about keeping the warning but making it optional instead?  That
> way, we could still use it in emacs.git.

Hm...  With a separate build target or something?  For a warning like
this to be useful, it has to actually trigger for most people when
building Emacs.

Or I guess the Makefile could just switch the warning on?

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

Re: master 977f3f27c5 2/2: Don't issue warnings for Unicode quotes for now

[Stefan Kangas]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> > How about keeping the warning but making it optional instead?  That
> > way, we could still use it in emacs.git.
>
> Hm...  With a separate build target or something?  For a warning like
> this to be useful, it has to actually trigger for most people when
> building Emacs.
>
> Or I guess the Makefile could just switch the warning on?

I was thinking of making it depend on a separate variable or user
option that we would set as a directory local variable.  Could such an
approach work?

Re: master 977f3f27c5 2/2: Don't issue warnings for Unicode quotes for now

[Lars Ingebrigtsen]

Stefan Kangas <stefan@marxist.se> writes:

> I was thinking of making it depend on a separate variable or user
> option that we would set as a directory local variable.  Could such an
> approach work?

Yes.  But it feels a bit awkward to introduce a variable just for this.

We do have byte-compile-warnings, which can be either t or a list, and
it would be nice if we could extend that to cover this use case, to
avoid introducing another mechanism.  But we want t to mean
"everything-but-this"...  Hm...

Perhaps we could make an explicit mapping: t to `(not
(docstrings-unicode-quotes ...))' and then introduce a new `emacs-build'
value that's the old t (basically)?  That way we could (in the future)
introduce more warnings just for the Emacs build that wouldn't affect
other usages of the byte compiler.

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

Re: master 977f3f27c5 2/2: Don't issue warnings for Unicode quotes for now

[Stefan Kangas]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Yes.  But it feels a bit awkward to introduce a variable just for this.

That's true.

> Perhaps we could make an explicit mapping: t to `(not
> (docstrings-unicode-quotes ...))' and then introduce a new `emacs-build'
> value that's the old t (basically)?  That way we could (in the future)
> introduce more warnings just for the Emacs build that wouldn't affect
> other usages of the byte compiler.

I like this approach.  It would allow us to introduce experimental
byte-compiler warnings: we could introduce new warnings more gently.
For example, perhaps the "long docstrings" warning could have started
out as experimental for a full major release to allow packages more
time to adapt.  (Not that it would have likely mattered much but
still.)

Re: master 977f3f27c5 2/2: Don't issue warnings for Unicode quotes for now

[Lars Ingebrigtsen]

Stefan Kangas <stefan@marxist.se> writes:

> I like this approach.  It would allow us to introduce experimental
> byte-compiler warnings: we could introduce new warnings more gently.
> For example, perhaps the "long docstrings" warning could have started
> out as experimental for a full major release to allow packages more
> time to adapt.

Yup.  So I've now implemented this in Emacs 29 (and reinstated the
Unicode quote check during the Emacs build).

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