Re: master 396355f: Re-fix charset issues when yanking non-plain-text elements

Re: master 396355f: Re-fix charset issues when yanking non-plain-text elements

[Eli Zaretskii]

>    (let ((data (gui-backend-get-selection (or type 'PRIMARY)
>                                           (or data-type 'STRING))))
>      (when (and (stringp data)
> -	       (setq data-type (get-text-property 0 'foreign-selection data)))
> +               ;; If this text property is set, then the data needs to
> +               ;; be decoded -- otherwise it has already been decoded
> +               ;; by the lower level functions.
> +               (get-text-property 0 'foreign-selection data))

The old code took the data-type from the value of the
foreign-selection property, but now we disregard that.  Is that on
purpose?

> +                    ;; Guess at the charset for types like text/html
> +                    ;; -- it can be anything, and different
> +                    ;; applications use different encodings.
> +                    ((string-match-p "\\`text/" (symbol-name data-type))
> +                     (decode-coding-string
> +                      data (car (detect-coding-string data))))

Did you really need the explicit call to detect-coding-string?  AFAIR,
decode-coding-string does that automatically if needed.  Did that fail
in some case?

Re: master 396355f: Re-fix charset issues when yanking non-plain-text elements

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> The old code took the data-type from the value of the
> foreign-selection property, but now we disregard that.  Is that on
> purpose?

Yes.  That made even binary data (like images) be mis-decoded.

Perhaps the logic here should be changed altogether -- and make all the
backend selection functions do the decoding themselves (since the
Windows one does it).

>> +                    ;; Guess at the charset for types like text/html
>> +                    ;; -- it can be anything, and different
>> +                    ;; applications use different encodings.
>> +                    ((string-match-p "\\`text/" (symbol-name data-type))
>> +                     (decode-coding-string
>> +                      data (car (detect-coding-string data))))
>
> Did you really need the explicit call to detect-coding-string?  AFAIR,
> decode-coding-string does that automatically if needed.  Did that fail
> in some case?

Can you give a nil to decode-coding-string as the coding system?  It's
not documented.

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

Re: master 396355f: Re-fix charset issues when yanking non-plain-text elements

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: emacs-devel@gnu.org
> Date: Thu, 11 Nov 2021 12:54:17 +0100
> 
> >> +                    ;; Guess at the charset for types like text/html
> >> +                    ;; -- it can be anything, and different
> >> +                    ;; applications use different encodings.
> >> +                    ((string-match-p "\\`text/" (symbol-name data-type))
> >> +                     (decode-coding-string
> >> +                      data (car (detect-coding-string data))))
> >
> > Did you really need the explicit call to detect-coding-string?  AFAIR,
> > decode-coding-string does that automatically if needed.  Did that fail
> > in some case?
> 
> Can you give a nil to decode-coding-string as the coding system?  It's
> not documented.

If by nil you means "don't know", then I think you should use
'undecided instead.  That's exactly what 'undecided means.

Re: master 396355f: Re-fix charset issues when yanking non-plain-text elements

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> If by nil you means "don't know", then I think you should use
> 'undecided instead.  That's exactly what 'undecided means.

That doesn't seem to be documented anywhere near this function (i.e.,
not in the doc string or in the "Explicit Encoding and Decoding" node in
the manual).

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

Re: master 396355f: Re-fix charset issues when yanking non-plain-text elements

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: emacs-devel@gnu.org
> Date: Fri, 12 Nov 2021 03:45:46 +0100
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > If by nil you means "don't know", then I think you should use
> > 'undecided instead.  That's exactly what 'undecided means.
> 
> That doesn't seem to be documented anywhere near this function (i.e.,
> not in the doc string or in the "Explicit Encoding and Decoding" node in
> the manual).

It does now.