bug#60841: 30.0.50; kill-ring-save pauses despite region being highlighted

["Kévin Le Gouguec" <kevin.legouguec@gmail.com>]

Eli Zaretskii <eliz@gnu.org> writes:

> Sorry, but I don't think I understand what you are proposing.

In the message you reply to, not much (and it was quite ingrate of me to
inflict that wall of text on this list, without the shadow of a
conclusion; apologies, will do better next time).

The message before that had a proposal, FWIW:

> So,
> 
> * :extend nil for both: they display differently (region will not be
>   extended, default will be),
> * :extend t for both: they display the same,
> * default has nil, region has t: they display the same,
> * default has t, region has nil: they display differently.
> 
> Ergo, assuming (a) I didn't mess something up (b) this is the expected
> behaviour, it seems that when considering :extend,
> (face-differs-from-default-p FACE FRAME) should check
> 
> (1) whether FACE's :extend is nil (regardless of default's :extend),
>     and
> (2) whether FACE's :underline or :background are "supported", as
>     reported by display-supports-face-attributes-p.

(
 Gregory then suggested that this example, from that same message…

   (set-face-background 'region (face-background 'default nil t))

 … is contrived, and shouldn't justify special-casing :extend in
 face-differs-from-default-p (well, shouldn't justify "adding too much
 complexity").

 To which I replied that the current code in indicate-copied-region was
 in answer to an actual user doing…

   (set-face-attribute 'region nil :background 'unspecified)

 … which runs afoul of the same problem as my contrived example: 'region
 remains "visually different" from 'default according to
 face-differs-from-default-p because of :extent, but for the purposes of
 indicate-copied-region, it _shouldn't_ be.
)

Now Gregory also suggested just adding an optional list of ignored
attributes to face-differs-from-default-p's parameters, so that
indicate-copied-region can set it to '(:extend) (IIUC; Gregory, let me
know if I've misunderstood).


And TBH I don't have a problem with this suggestion: it punts to the
caller, who can do TRT; it keeps face-differs-from-default-p relatively
simple, we can go with that.  My only objection is that I can't imagine
a caller ever wanting to _not_ pass '(:extend), but I admit that my own
"proposition" sounds convoluted.


(Hm, and against my better judgement I went ahead and compared
gui_supports_face_attributes_p vs tty_supports_face_attributes_p, and I
see that they handle :extend differently and *mashes C-c C-c with
forehead before fingers can type another wall of text*)