bug#12890: read-face-attribute, face-valid-attribute-values do not handle all :underline etc forms

bug#12890: read-face-attribute, face-valid-attribute-values do not handle all :underline etc forms

[Glenn Morris]

Package: emacs
Version: 24.2.50

You can't use read-face-attribute to set list values for :box, :stipple,
or :underline:

emacs -Q --eval "(read-face-attribute 'underline :underline)"
Start to enter: (:color "red")
and get stopped with "no match" when you type the space.
If you force it with C-q space, it takes your input and returns it, but
silently does not actually set the :underline attribute to it.

This is because face-valid-attribute-values does not return such values.

This affects eg M-x set-face-underline.

bug#12890: read-face-attribute, face-valid-attribute-values do not handle all :underline etc forms

[Lars Ingebrigtsen]

Glenn Morris <rgm@gnu.org> writes:

> You can't use read-face-attribute to set list values for :box, :stipple,
> or :underline:
>
> emacs -Q --eval "(read-face-attribute 'underline :underline)"
> Start to enter: (:color "red")
> and get stopped with "no match" when you type the space.
> If you force it with C-q space, it takes your input and returns it, but
> silently does not actually set the :underline attribute to it.
>
> This is because face-valid-attribute-values does not return such values.
>
> This affects eg M-x set-face-underline.

I tried reproducing this in Emacs 29.  It does indeed still need `C-q'
to enter the space, but after doing that, it seems to actually do the
rest correctly.

I did:

M-x set-face-underline RET
default RET
(:colorC-q "red") RET

and the default face got a red underline.

The `C-q' should be fixed, but do you also see the same behaviour for
the rest?

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

bug#12890: read-face-attribute, face-valid-attribute-values do not handle all :underline etc forms

[Lars Ingebrigtsen]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> I tried reproducing this in Emacs 29.  It does indeed still need `C-q'
> to enter the space, but after doing that, it seems to actually do the
> rest correctly.
>
> I did:
>
> M-x set-face-underline RET
> default RET
> (:colorC-q "red") RET
>
> and the default face got a red underline.
>
> The `C-q' should be fixed, but do you also see the same behaviour for
> the rest?

Actually, I have no idea how to fix the `C-q' here.  We do want
completion because we want to complete over colour names, and (by
default) `SPC' does completion.

It might be nice to have the `SPC' in minibuffer-local-completion-map to
be more DWIM -- don't try to complete if it's "obvious" that the user
doesn't want completion.  Myself, I've disabled completion on `SPC'
because it's really annoying, but perhaps there is something we could do
here?  I.e., if there are no completions, then just insert a `SPC' and
make that the new default binding for `SPC' here?  

Because I think that would make usability here a lot better.  (It could
say "No completions; inserting space".)

Opinions?

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

bug#12890: read-face-attribute, face-valid-attribute-values do not handle all :underline etc forms

[Lars Ingebrigtsen]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> It might be nice to have the `SPC' in minibuffer-local-completion-map to
> be more DWIM -- don't try to complete if it's "obvious" that the user
> doesn't want completion.  Myself, I've disabled completion on `SPC'
> because it's really annoying, but perhaps there is something we could do
> here?  I.e., if there are no completions, then just insert a `SPC' and
> make that the new default binding for `SPC' here?  
>
> Because I think that would make usability here a lot better.  (It could
> say "No completions; inserting space".)
>
> Opinions?

Thinking about this a bit more, I think it's likely that there are
people that are relying on SPC being bound to
`minibuffer-complete-word', and it would be too annoying to change this
behaviour (even when there are no completions to be had).

It could be added as an option, but I don't think it would be a very
popular option -- people that don't like the SPC behaviour already have

(define-key minibuffer-local-completion-map " " 'self-insert-command)

and the rest wouldn't want to use this, anyway.  So I'm closing this bug
report.

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