Should minibuffer prompt be made intangible by default? (Was debbugs 21874: 25.0.50; point-entered no longer works)

Should minibuffer prompt be made intangible by default? (Was debbugs 21874: 25.0.50; point-entered no longer works)

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 1466 bytes --]

Hi Stefan,

I agree with what David has to say; I too believe that there are a lot more
users expecting the minibuffer prompt to be intangible.

For now, I need to add the below to my emacs config (as I don't set the
customize variables using the GUI approach, and I would like to simply
append stuff to whatever the default value is):

(let (;; (get ..)                   -> ((quote (read-only t face
minibuffer-prompt)))
      ;; (car (get ..))             -> (quote (read-only t face
minibuffer-prompt))
      ;; (cdr (car (get ..)))       -> ((read-only t face
minibuffer-prompt))
      ;; (car (cdr (car (get ..)))) -> (read-only t face minibuffer-prompt)
      (default (car (cdr (car (get 'minibuffer-prompt-properties
'standard-value)))))
      (dont-touch-prompt-prop '(cursor-intangible t)))
  ;; When `cursor-intangible' property is detected in
`minibuffer-prompt-properties',
  ;; `cursor-intangible-mode' is automatically added to
`minibuffer-setup-hook'
  ;; (see cus-start.el).
  (custom-set-variables '(minibuffer-prompt-properties
                          (append default dont-touch-prompt-prop)
                          nil nil "Make the minibuffer prompt
intangible.")))

It is helpful that the minibuffer-setup-hook is updated automatically. But
I would also vote for '(cursor-intangible t) to be part of the default
value of minibuffer-prompt-properties. That way majority of the users would
see the minibuffer prompt behaving as they expect.

[-- Attachment #2: Type: text/html, Size: 2753 bytes --]

Re: Should minibuffer prompt be made intangible by default? (Was debbugs 21874: 25.0.50; point-entered no longer works)

[Stefan Monnier]

>       ;; (car (cdr (car (get ..)))) -> (read-only t face minibuffer-prompt)
            ^^^^^^^^
This "car(cdr" should be spelled "eval" (there's no guarantee that the
arg has always the shape (quote ...something..)).

>   (custom-set-variables '(minibuffer-prompt-properties

Don't do that.  There should only ever be one call to
custom-set-variables, auto-written by Custom itself.  As soon as you
move away from that, you're entering dangerous territory.

You should probably use customize-set-variable instead.

Tho I'd personally recommend that if you don't want to use the Custom
UI, then don't use Custom from Elisp either, and just use straight:

    (setq minibuffer-prompt-properties
          (append ... minibuffer-prompt-properties))
    (add-hook 'minibuffer-setup-hook #'cursor-sensor-mode)


-- Stefan

Re: Should minibuffer prompt be made intangible by default? (Was debbugs 21874: 25.0.50; point-entered no longer works)

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 1199 bytes --]

On Thu, Mar 31, 2016 at 8:38 AM, Stefan Monnier <monnier@iro.umontreal.ca>
wrote:

> >       ;; (car (cdr (car (get ..)))) -> (read-only t face
> minibuffer-prompt)
>             ^^^^^^^^
> This "car(cdr" should be spelled "eval" (there's no guarantee that the
> arg has always the shape (quote ...something..)).
>

Thank you. That looks much saner too :)


>
> >   (custom-set-variables '(minibuffer-prompt-properties
>
> Don't do that.  There should only ever be one call to
> custom-set-variables, auto-written by Custom itself.  As soon as you
> move away from that, you're entering dangerous territory.
>

That was an oversight on my part. I realized the problem practically later
on and ended up having what you mentioned below; just that I add
cursor-intangible-mode to the hook instead of cursor-sensor-mode.


> You should probably use customize-set-variable instead.
>
> Tho I'd personally recommend that if you don't want to use the Custom
> UI, then don't use Custom from Elisp either, and just use straight:
>
>     (setq minibuffer-prompt-properties
>           (append ... minibuffer-prompt-properties))
>     (add-hook 'minibuffer-setup-hook #'cursor-sensor-mode)



--
Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 2028 bytes --]

Re: Should minibuffer prompt be made intangible by default? (Was debbugs 21874: 25.0.50; point-entered no longer works)

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 499 bytes --]

On 03/31/2016 02:38 PM, Stefan Monnier wrote:
> Tho I'd personally recommend that if you don't want to use the Custom
> UI, then don't use Custom from Elisp either, and just use straight:
> 
>     (setq minibuffer-prompt-properties
>           (append ... minibuffer-prompt-properties))
>     (add-hook 'minibuffer-setup-hook #'cursor-sensor-mode)

Is that right? That will work in this case, but in general isn't it a bad idea to setq a defcustom, given that it won't run its :set form?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Should minibuffer prompt be made intangible by default? (Was debbugs 21874: 25.0.50; point-entered no longer works)

[Stefan Monnier]

> Is that right? That will work in this case, but in general isn't it a bad
> idea to setq a defcustom, given that it won't run its :set form?

That's why there's the

   (add-hook 'minibuffer-setup-hook #'cursor-intangible-mode)

You should always be able to set things up with good ol' setq
and friends.  Custom is something you *can* use but shouldn't have to
(tho, admittedly, currently our story for Elisp-level customization of
faces is rather poor, so for those I think that Custom is the only
reasonable option).


        Stefan

Re: Should minibuffer prompt be made intangible by default? (Was debbugs 21874: 25.0.50; point-entered no longer works)

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 553 bytes --]

On 03/31/2016 11:37 PM, Stefan Monnier wrote:
> (tho, admittedly, currently our story for Elisp-level customization of
> faces is rather poor, so for those I think that Custom is the only
> reasonable option).

Thanks! Faces are one of the thing I *don't* use custom for. AFAIK it does not let you create and manipulate fontsets; I don't think we have any way to specify these properly, in fact. At least, the question at https://emacs.stackexchange.com/questions/17205/how-do-i-set-up-font-fallback-in-a-robust-way didn't get a single answer. 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Alexis]

Clément Pit--Claudel <clement.pit@gmail.com> writes:

> Thanks! Faces are one of the thing I *don't* use custom 
> for. AFAIK it does not let you create and manipulate fontsets; I 
> don't think we have any way to specify these properly, in 
> fact. At least, the question at 
> https://emacs.stackexchange.com/questions/17205/how-do-i-set-up-font-fallback-in-a-robust-way 
> didn't get a single answer.

i know i'm sounding like a broken record, but the `unicode-fonts` 
might be what you're looking for:

http://melpa.org/#/unicode-fonts


Alexis.

Re: Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Eli Zaretskii]

> From: Alexis <flexibeast@gmail.com>
> Date: Fri, 01 Apr 2016 12:40:35 +1100
> 
> > Thanks! Faces are one of the thing I *don't* use custom 
> > for. AFAIK it does not let you create and manipulate fontsets; I 
> > don't think we have any way to specify these properly, in 
> > fact. At least, the question at 
> > https://emacs.stackexchange.com/questions/17205/how-do-i-set-up-font-fallback-in-a-robust-way 
> > didn't get a single answer.
> 
> i know i'm sounding like a broken record, but the `unicode-fonts` 
> might be what you're looking for:
> 
> http://melpa.org/#/unicode-fonts

Installing fonts doesn't necessarily resolve the fontset configuration
problems, especially since no package can know what other fonts are
available on the end-user system.

So I'm not sure this has any relation to the question on
stackexchange, because that question was describing a very specific
situation where other fonts were involved (in unspecified ways).

(The reason that question got no answers is because it was asked in
the wrong place, btw.)

Re: Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 183 bytes --]

On 04/01/2016 09:27 AM, Eli Zaretskii wrote:
> (The reason that question got no answers is because it was asked in
> the wrong place, btw.)

Do you think I should ask it here?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Eli Zaretskii]

> From: Clément Pit--Claudel <clement.pit@gmail.com>
> Date: Fri, 1 Apr 2016 12:00:54 +0200
> 
> On 04/01/2016 09:27 AM, Eli Zaretskii wrote:
> > (The reason that question got no answers is because it was asked in
> > the wrong place, btw.)
> 
> Do you think I should ask it here?

Here or at least on help-gnu-emacs, yes.  Font setup in Emacs is an
obscure topic, and I don't think anyone of those who know something
about that read stackexchange.

Re: Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Alexis]

Eli Zaretskii <eliz@gnu.org> writes:

> Installing fonts doesn't necessarily resolve the fontset 
> configuration problems, especially since no package can know 
> what other fonts are available on the end-user system.

Sorry, i'm not sure you understand what the `unicode-fonts` 
package does? It doesn't provide fonts itself, but configures 
fontsets to use fonts on a users' system for maximum Unicode 
coverage. As the package's README.md says:

    font mappings via fontsets are a bit difficult to 
    configure. In addition, the default setup does not always pick 
    the most legible fonts. As the manual warns, the choice of 
    font actually displayed for a non-ASCII character is "somewhat 
    random". 
 
    The Unicode standard provides a way to organize font mappings: 
    it divides character ranges into logical groups called 
    "blocks". This library configures Emacs in a Unicode-friendly 
    way by providing mappings from 
 
        each Unicode block  ---to--->   a font with good coverage 
 
    and makes the settings available via the customization 
    interface. 
 
    This library provides font mappings for 233 of the 255 blocks 
    in the Unicode 8.0 standard which are public and have 
    displayable characters. It assumes that 6 Latin blocks are 
    covered by the default font. 16/255 blocks are not mapped to 
    any known font.

(However, the README.md /does/ suggests some fonts for users to 
download to facilitate the `unicode-fonts` package's work.)


Alexis.

Re: Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Eli Zaretskii]

> From: Alexis <flexibeast@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Sat, 02 Apr 2016 19:57:45 +1100
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Installing fonts doesn't necessarily resolve the fontset 
> > configuration problems, especially since no package can know 
> > what other fonts are available on the end-user system.
> 
> Sorry, i'm not sure you understand what the `unicode-fonts` 
> package does?

I've read the code, but that doesn't necessarily mean I didn't miss
anything, of course.

However, I did read README.md as well, and upon re-reading now the
text you posted, I find nothing in it that would contradict what I
said.

> It doesn't provide fonts itself, but configures fontsets to use
> fonts on a users' system for maximum Unicode coverage.

I know; I've read the code before that original reply.

Maybe we are talking about 2 different issues.  My understanding of
the original stackexchange question was that the problems there were
related to the fact that fontset setup caused some characters to be
displayed with fonts other than what the user saw before changing his
setup.  It should be clear that the fonts used before changing the
setup depend on what is installed on the user's system, something that
no package can take into consideration.  Thus this part of what I
wrote:

> > Installing fonts doesn't necessarily resolve the fontset 
> > configuration problems, especially since no package can know
> >                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> > what other fonts are available on the end-user system.
> > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

IOW, the problem here that I referred to is this:

  . User complains that some characters cannot be displayed
  . User is advised to install a fontset setup and additional fonts to
    fix that
  . The fontset setup causes some characters previously displayed with
    font A to be now displayed with font B, and the user doesn't like that

My point is that, once you get to the 3rd item, no external package
can be the correct answer; instead, you need to teach the user to
configure his/her fontset according to his/her preferences.  In
particular, the fontset setup should be based not only on Unicode
blocks, but also on the capabilities of the fonts present in the
original system configuration, and the user's preferences for using
particular font(s) for displaying particular ranges of characters.

Re: Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Alexis]

Eli Zaretskii <eliz@gnu.org> writes:
 
>> > Installing fonts doesn't necessarily resolve the fontset 
>> > configuration problems, especially since no package can know 
>> > what other fonts are available on the end-user system.
>> 
>> Sorry, i'm not sure you understand what the `unicode-fonts` 
>> package does?
>
> I've read the code, but that doesn't necessarily mean I didn't 
> miss anything, of course.
>
> However, I did read README.md as well, and upon re-reading now 
> the text you posted, I find nothing in it that would contradict 
> what I said.

You wrote "Installing fonts doesn't ...", and didn't write 
something like "Installing fonts and then configuring them using 
that package doesn't .... "Since `unicode-fonts` doesn't actually 
install fonts, i think it wasn't entirely unreasonable of me to 
assume that you thought that the package merely installed fonts.

> My point is that, once you get to the 3rd item, no external 
> package can be the correct answer; instead, you need to teach 
> the user to configure his/her fontset according to his/her 
> preferences.  In particular, the fontset setup should be based 
> not only on Unicode blocks, but also on the capabilities of the 
> fonts present in the original system configuration, and the 
> user's preferences for using particular font(s) for displaying 
> particular ranges of characters.

Sure. On my system, `unicode-fonts` actually decided to use glyphs 
from the "Linux Biolinum" family for displaying Hebrew, which i 
found basically illegible. So my setup manually prioritises Ezra 
SIL:

    (setcdr (assoc "Hebrew" unicode-fonts-block-font-mapping) 
            '(("Ezra SIL" "Ezra SIL SR" "Arial Hebrew" "Raanana" 
            "New Peninim MT" "Aharoni" "David" "FrankRuehl" 
            "Gisha" "Levenim MT" "Narkisim" "Rod" "Cardo" "Courier 
            New" "Adobe Hebrew" "Code2000" "Aramaic Imperial Yeb" 
            "Microsoft Sans Serif" "Tahoma" "Lucida Sans Unicode" 
            "Arial Unicode MS" "Arial" "Quivira" "Everson 
            Mono:weight=boeld" "ALPHABETUM Unicode"))) 

At any rate, i've clearly misunderstood the problem - sorry! - so 
i'll bow out now.


Alexis.

Re: Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 787 bytes --]

On 04/01/2016 03:40 AM, Alexis wrote:
> Clément Pit--Claudel <clement.pit@gmail.com> writes:
> 
>> Thanks! Faces are one of the thing I *don't* use custom for. AFAIK
>> it does not let you create and manipulate fontsets; I don't think
>> we have any way to specify these properly, in fact. At least, the
>> question at
>> https://emacs.stackexchange.com/questions/17205/how-do-i-set-up-font-fallback-in-a-robust-way
>> didn't get a single answer.
> 
> i know i'm sounding like a broken record, but the `unicode-fonts`
> might be what you're looking for:

No, I don't think so. I don't think unicode-fonts provides a good way to configure font-fallback; instead, it just comes with a large set of good defaults, but this is not what that question was about.

Clément.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Configuring fontsets with `unicode-fonts` [Was: Re: Should minibuffer prompt be made intangible by default?]

[Alexis]

Clément Pit--Claudel <clement.pit@gmail.com> writes:

> No, I don't think so. I don't think unicode-fonts provides a 
> good way to configure font-fallback; instead, it just comes with 
> a large set of good defaults, but this is not what that question 
> was about.

Okay. i'm sorry for misunderstanding the problem, and for the 
noise.


Alexis.