Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Robert Pluim]

>>>>> On Thu, 16 Sep 2021 03:38:53 -0400 (EDT), stefankangas@gmail.com (Stefan Kangas) said:
    Stefan>     * src/fns.c (Fyes_or_no_p): Doc fix to reflect that a trailing space
    Stefan>     is no longer needed; one is added or removed automatically.

Hi Stefan,

I just tested this on master, and I donʼt think this is qute accurate:

- `y-or-n-p' will append a space if there is one
- it wonʼt remove one
- `yes-or-no-p' never modifies the passed prompt

Robert
-- 

Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Stefan Kangas]

Robert Pluim <rpluim@gmail.com> writes:

>>>>>> On Thu, 16 Sep 2021 03:38:53 -0400 (EDT), stefankangas@gmail.com (Stefan Kangas) said:
>     Stefan>     * src/fns.c (Fyes_or_no_p): Doc fix to reflect that a trailing space
>     Stefan>     is no longer needed; one is added or removed automatically.
>
> I just tested this on master, and I donʼt think this is qute accurate:

Hmm, I guess you're right.  Not sure what I was smoking:

> - `y-or-n-p' will append a space if there is one
> - it wonʼt remove one

    (y-or-n-p "foo")     => prompt is "foo (y or n) "
    (y-or-n-p "foo ")    => prompt is "foo (y or n) "

> - `yes-or-no-p' never modifies the passed prompt

    (yes-or-no-p "foo")  => prompt is "foo(yes or no) "
    (yes-or-no-p "foo ") => prompt is "foo (yes or no) "

I think the best fix here is to change `yes-or-no-p' to work the same
way as `y-or-n-p'.

Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Robert Pluim]

>>>>> On Tue, 3 Jan 2023 15:33:21 +0000, Stefan Kangas <stefankangas@gmail.com> said:

    Stefan> Robert Pluim <rpluim@gmail.com> writes:
    >>>>>>> On Thu, 16 Sep 2021 03:38:53 -0400 (EDT), stefankangas@gmail.com (Stefan Kangas) said:
    Stefan> * src/fns.c (Fyes_or_no_p): Doc fix to reflect that a trailing space
    Stefan> is no longer needed; one is added or removed automatically.
    >> 
    >> I just tested this on master, and I donʼt think this is qute accurate:

    Stefan> Hmm, I guess you're right.  Not sure what I was smoking:

    >> - `y-or-n-p' will append a space if there is one

*isnʼt* one, I meant. This stuff is hard :-)

    >> - it wonʼt remove one

    Stefan>     (y-or-n-p "foo")     => prompt is "foo (y or n) "
    Stefan>     (y-or-n-p "foo ")    => prompt is "foo (y or n) "

I was thinking (y-or-n-p "foo  ")    => prompt is "foo  (y or n)"

which we could consider fixing (but itʼs long-standing behaviour)

    >> - `yes-or-no-p' never modifies the passed prompt

    Stefan>     (yes-or-no-p "foo")  => prompt is "foo(yes or no) "
    Stefan>     (yes-or-no-p "foo ") => prompt is "foo (yes or no) "

    Stefan> I think the best fix here is to change `yes-or-no-p' to work the same
    Stefan> way as `y-or-n-p'.

ENOPATCH 😜

(and as someone pointed out elsewhere `y-or-n-p' and `yes-or-no-p'
could do with some refactoring.

Robert
-- 

Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Jean Louis]

* Stefan Kangas <stefankangas@gmail.com> [2023-01-03 18:35]:
> Robert Pluim <rpluim@gmail.com> writes:
> 
> >>>>>> On Thu, 16 Sep 2021 03:38:53 -0400 (EDT), stefankangas@gmail.com (Stefan Kangas) said:
> >     Stefan>     * src/fns.c (Fyes_or_no_p): Doc fix to reflect that a trailing space
> >     Stefan>     is no longer needed; one is added or removed automatically.
> >
> > I just tested this on master, and I donʼt think this is qute accurate:
> 
> Hmm, I guess you're right.  Not sure what I was smoking:
> 
> > - `y-or-n-p' will append a space if there is one
> > - it wonʼt remove one
> 
>     (y-or-n-p "foo")     => prompt is "foo (y or n) "
>     (y-or-n-p "foo ")    => prompt is "foo (y or n) "
> 
> > - `yes-or-no-p' never modifies the passed prompt
> 
>     (yes-or-no-p "foo")  => prompt is "foo(yes or no) "
>     (yes-or-no-p "foo ") => prompt is "foo (yes or no) "
> 
> I think the best fix here is to change `yes-or-no-p' to work the same
> way as `y-or-n-p'.

At least one space shall be there, but please don't remove trailing
spaces, as that is users' choice, so that following unusual example
may work:

(y-or-n-p "                               Are you sure?             ")



-- 
Jean

Take action in Free Software Foundation campaigns:
https://www.fsf.org/campaigns

In support of Richard M. Stallman
https://stallmansupport.org/

Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Robert Pluim]

>>>>> On Wed, 4 Jan 2023 11:43:41 +0300, Jean Louis <bugs@gnu.support> said:

    Jean> * Stefan Kangas <stefankangas@gmail.com> [2023-01-03 18:35]:
    >> Robert Pluim <rpluim@gmail.com> writes:
    >> 
    >> >>>>>> On Thu, 16 Sep 2021 03:38:53 -0400 (EDT), stefankangas@gmail.com (Stefan Kangas) said:
    >> >     Stefan>     * src/fns.c (Fyes_or_no_p): Doc fix to reflect that a trailing space
    >> >     Stefan>     is no longer needed; one is added or removed automatically.
    >> >
    >> > I just tested this on master, and I donʼt think this is qute accurate:
    >> 
    >> Hmm, I guess you're right.  Not sure what I was smoking:
    >> 
    >> > - `y-or-n-p' will append a space if there is one
    >> > - it wonʼt remove one
    >> 
    >> (y-or-n-p "foo")     => prompt is "foo (y or n) "
    >> (y-or-n-p "foo ")    => prompt is "foo (y or n) "
    >> 
    >> > - `yes-or-no-p' never modifies the passed prompt
    >> 
    >> (yes-or-no-p "foo")  => prompt is "foo(yes or no) "
    >> (yes-or-no-p "foo ") => prompt is "foo (yes or no) "
    >> 
    >> I think the best fix here is to change `yes-or-no-p' to work the same
    >> way as `y-or-n-p'.

    Jean> At least one space shall be there, but please don't remove trailing
    Jean> spaces, as that is users' choice, so that following unusual example
    Jean> may work:

    Jean> (y-or-n-p "                               Are you sure?             ")

You mean 'yes-or-no-p', right? 😺

Thatʼs the historical behaviour, I donʼt see a strong case for
changing it (but `yes-or-no-p' should be fixed to add a space if there
isnʼt one).

Robert
-- 

RE: [External] : Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Drew Adams]

> (but `yes-or-no-p' should be fixed to add a space if there
> isnʼt one).

I'm not following this thread, so apologies if
my comment is irrelevant or misguided.

I disagree that functions such as `yes-or-no-p'
should mess with the prompt.  If a user wants
a space after the colon, and the user forgets
to add one, that's on the user.

If a user wants NO space after the colon for
some reason, however misguided the author of
`yes-or-no-p' might think that choice is, it
should be respected - there's no reason _not_
to respect it.

`yes-or-no-p' and the like shouldn't try to
second-guess users.  Don't let it suffer from
"dwim" lameness, please.  It's up to users to
provide the prompt text they want - they have
tons of ways to do that, including `format'.

Pretty much all of the prompting functions
that append a space, or a colon plus a space,
are misguided.  OK, there's legacy, and we
accommodate that.  But let's please not ADD
to the problem by introducing more examples
of it.  Not willingly and knowingly, at any
rate.

Again, I'm not following this thread.  But
even the Subject line bit "trailing space is
no longer needed" makes me cringe, at least
in my ignorance of what "fix" is intended.

Lots of what exists might be considered "no
longer needed" by someone.  Whether users
already make use of it shouldn't be ignored.

Re: [External] : Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Robert Pluim]

>>>>> On Wed, 4 Jan 2023 15:53:59 +0000, Drew Adams <drew.adams@oracle.com> said:

    >> (but `yes-or-no-p' should be fixed to add a space if there
    >> isnʼt one).

    Drew> I'm not following this thread, so apologies if
    Drew> my comment is irrelevant or misguided.

If you were, your comments would have been shorter

- `y-or-n-p' has, since emacs 24.3 or so, appended a space to the prompt
if there wasnʼt one
- `yes-or-no-p' doesnʼt mess with the prompt

This discussion is about the documentation of those two facts, which
needs adjusting since itʼs currently incorrect. There was a suggestion
by me to maybe align `yes-or-no-p' behaviour (what you call "dwim") to
match `y-or-n-p', but since thatʼs 'legacy' I doubt itʼs going to be
done.

Robert
-- 

RE: [External] : Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Drew Adams]

>     >> (but `yes-or-no-p' should be fixed to add a space if there
>     >> isnʼt one).
> 
>     Drew> I'm not following this thread, so apologies if
>     Drew> my comment is irrelevant or misguided.
> 
> If you were, your comments would have been shorter

If I were what?

> - `y-or-n-p' has, since emacs 24.3 or so, appended a space to the prompt
>   if there wasnʼt one
> - `yes-or-no-p' doesnʼt mess with the prompt
> 
> This discussion is about the documentation of those two facts, which
> needs adjusting since itʼs currently incorrect. 

Oh, it's the _doc_ that needs adjusting to those facts?
OK, then.  So no change in the _behavior_ is being
considered.

> There was a suggestion by me to maybe align `yes-or-no-p'
> behaviour (what you call "dwim") to match `y-or-n-p',
> but since thatʼs 'legacy' I doubt itʼs going to be done.

OK, so nothing is changing, apparently.  Sorry for the noise.

Re: [External] : Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Robert Pluim]

>>>>> On Wed, 4 Jan 2023 16:12:23 +0000, Drew Adams <drew.adams@oracle.com> said:

    >> >> (but `yes-or-no-p' should be fixed to add a space if there
    >> >> isnʼt one).
    >> 
    Drew> I'm not following this thread, so apologies if
    Drew> my comment is irrelevant or misguided.
    >> 
    >> If you were, your comments would have been shorter

    Drew> If I were what?

following the thread

    >> - `y-or-n-p' has, since emacs 24.3 or so, appended a space to the prompt
    >> if there wasnʼt one
    >> - `yes-or-no-p' doesnʼt mess with the prompt
    >> 
    >> This discussion is about the documentation of those two facts, which
    >> needs adjusting since itʼs currently incorrect. 

    Drew> Oh, it's the _doc_ that needs adjusting to those facts?
    Drew> OK, then.  So no change in the _behavior_ is being
    Drew> considered.

Not seriously, no

    >> There was a suggestion by me to maybe align `yes-or-no-p'
    >> behaviour (what you call "dwim") to match `y-or-n-p',
    >> but since thatʼs 'legacy' I doubt itʼs going to be done.

    Drew> OK, so nothing is changing, apparently.  Sorry for the noise.

See, this is why you should follow the thread :-)

Robert
-- 

Re: [External] : Re: master 7f53446: Doc fix for y-or-n-p; trailing space is no longer needed

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Cc: Stefan Kangas <stefankangas@gmail.com>,  "emacs-devel@gnu.org"
>  <emacs-devel@gnu.org>
> Date: Wed, 04 Jan 2023 17:07:37 +0100
> 
> - `y-or-n-p' has, since emacs 24.3 or so, appended a space to the prompt
> if there wasnʼt one
> - `yes-or-no-p' doesnʼt mess with the prompt
> 
> This discussion is about the documentation of those two facts, which
> needs adjusting since itʼs currently incorrect. There was a suggestion
> by me to maybe align `yes-or-no-p' behaviour (what you call "dwim") to
> match `y-or-n-p', but since thatʼs 'legacy' I doubt itʼs going to be
> done.

I don't see any reasons whatsoever to mention this DWIM-ish behavior
in a doc string.  We made these functions behave like that because we
believe that would silently DTRT in every reasonable use case, so
unless someone complains about this behavior, and documenting it
somehow satisfies the complaint, we should not confuse Lisp
programmers with these details, about which they should not care.

So I've went ahead and removed this detail from the two doc string,
including the fact that y-or-n-p adds a space _before_ the
parentheses.  I hope this will put this thread to rest.