Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

[-- Attachment #1: Type: text/plain, Size: 633 bytes --]

I have some older code with a lot of calls to (interactive-p) for which
I am having to substitute (called-interactively-p 'interactive) to get the
same behavior with a non-obsoleted call.  Visually, this does not read well
to me and I imagine others.  What do people think of making the argument
optional and defaulting it to the behavior of (interactive-p) which I
believe equates to the 'interactive argument value?

Then we would have: if (called-interactively-p) ...) in a lot of places and
that would look much better and be much easier to call properly.

When more complex behavior is needed, the argument could be given.

Bob

[-- Attachment #2: Type: text/html, Size: 1331 bytes --]

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[John Wiegley]

[-- Attachment #1: Type: text/plain, Size: 746 bytes --]

>>>>> Robert Weiner <rsw@gnu.org> writes:

> I have some older code with a lot of calls to (interactive-p) for which I am
> having to substitute (called-interactively-p 'interactive) to get the same
> behavior with a non-obsoleted call. Visually, this does not read well to me
> and I imagine others. What do people think of making the argument optional
> and defaulting it to the behavior of (interactive-p) which I believe equates
> to the 'interactive argument value?

I like this suggestion; especially since (called-interactively-p 'interactive)
seems needlessly repetitive.

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 629 bytes --]

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

[-- Attachment #1: Type: text/plain, Size: 281 bytes --]

On Tue, Jun 28, 2016 at 3:09 PM, John Wiegley <jwiegley@gmail.com> wrote:

> I like this suggestion; especially since (called-interactively-p
> 'interactive)
> seems needlessly repetitive.
>

​Please make it so, John.  As Forrest Gump said, "One less thing."

Bob
​

[-- Attachment #2: Type: text/html, Size: 924 bytes --]

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Andreas Röhler]

On 28.06.2016 21:09, John Wiegley wrote:
>>>>>> Robert Weiner <rsw@gnu.org> writes:
>> I have some older code with a lot of calls to (interactive-p) for which I am
>> having to substitute (called-interactively-p 'interactive) to get the same
>> behavior with a non-obsoleted call. Visually, this does not read well to me
>> and I imagine others. What do people think of making the argument optional
>> and defaulting it to the behavior of (interactive-p) which I believe equates
>> to the 'interactive argument value?
> I like this suggestion; especially since (called-interactively-p 'interactive)
> seems needlessly repetitive.
>

What about re-honoring good and simple interactive-p?

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Glenn Morris]

It's easy to search for prior discussion on this. Eg

http://lists.gnu.org/archive/html/emacs-devel/2015-07/msg00326.html

   That would defeat the whole purpose (which is to force the author
   to make a choice between the two kinds, rather than to use one without
   knowing that there's a choice).

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

[-- Attachment #1: Type: text/plain, Size: 1068 bytes --]

On Tue, Jun 28, 2016 at 4:29 PM, Glenn Morris <rgm+news@gnu.org> wrote:

>
> It's easy to search for prior discussion on this. Eg
>
> http://lists.gnu.org/archive/html/emacs-devel/2015-07/msg00326.html
>
>    That would defeat the whole purpose (which is to force the author
>    to make a choice between the two kinds, rather than to use one without
>    knowing that there's a choice).
>

​This is a frequently called function and it is not good design to make
everyone who has encountered it and read the documentation forever more
have to add that argument when so many of the use cases would not require
it.  And again, visually, it looks bad and is much more complex than the
prior (interactive-p).  Yes, there will be newbies who do not know there is
an argument that they could provide because they just copied a snippet of
code but that is far from a major issue and not much of a justification to
create more work for the bulk of developers.

Does anyone concur or disagree?

I will have a look at the prior discussion.

Bob
​

[-- Attachment #2: Type: text/html, Size: 2506 bytes --]

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

[-- Attachment #1: Type: text/plain, Size: 968 bytes --]

Much of the prior discussion on this topic seems to favor my suggestion and
even suggest that people would like (interactive-p) non-deprecated.  -- Bob

Here is a taste:

*From*: Drew Adams
*Subject*: RE: Brittleness of called-interactively-p
*Date*: Thu, 16 Jul 2015 10:36:41 -0700 (PDT)

FWIW, I still use `interactive-p' in much of my code, because the code
needs to work also with older Emacs versions.  And because I have never
noticed any problem, for this code anyway, with `interactive-p'.

What's more, the doc for `interactive-p' does not really tell you how
to replace it - it just says to use `called-interactively-p', without
any mention of which argument gives you the behavior you had previously
with `interactive-p' or similar-but-somehow-improved behavior.

`interactive-p' was indeed used heavily, over decades.  And it is no
doubt still in use quite a bit.  Too bad there is next-to-no guidance
on how to use `called-interactively-p' to replace it.

[-- Attachment #2: Type: text/html, Size: 1784 bytes --]

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

  > ​This is a frequently called function and it is not good design to make
  > everyone who has encountered it and read the documentation forever more
  > have to add that argument when so many of the use cases would not require
  > it.

Maybe there is bad design in this -- in the fact that programmers need
to make this complex choice to avoid giving users bad behavior.  Good
design would, I suppose, make the choice unnecessary.

Telling people to ignore the choice, or choose at random, while
the choice is still necessary, would not be better design.
-- 
Dr Richard Stallman
President, Free Software Foundation (gnu.org, fsf.org)
Internet Hall-of-Famer (internethalloffame.org)
Skype: No way! See stallman.org/skype.html.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

On Wed, Jun 29, 2016 at 4:02 PM, Richard Stallman <rms@gnu.org> wrote:
> Telling people to ignore the choice, or choose at random, while
> the choice is still necessary, would not be better design.

That is an argument for never allowing optional parameters to
functions.  But I would assume you are not against these entirely.
Any time you make an argument optional, the function caller still has
the function signature and documentation to rely on when deciding how
to make the call.  The only difference is he has the power to avoid
sending the argument when he need not.

There was a function, interactive-p, that many people used without any
arguments and found that useful.  Now (called-interactively-p
'interactive) is equivalent but makes code much uglier (I know as I
just changed all my code to use it and it is much less readable now).
The request is for a function of no arguments that does what
interactive-p did; it could be called-interactively-p with an optional
argument signature or it could be another function that provides an
equivalent.

Does that affect your thinking?

Bob

RE: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Drew Adams]

> There was a function, interactive-p, that many people used without any
> arguments and found that useful.  Now (called-interactively-p
> 'interactive) is equivalent but makes code much uglier (I know as I
> just changed all my code to use it and it is much less readable now).
> The request is for a function of no arguments that does what
> interactive-p did; it could be called-interactively-p with an optional
> argument signature or it could be another function that provides an
> equivalent.

If (called-interactively-p 'interactive) is equivalent to
`interactive-p', and if "the call is for a function of no
arguments that does what `interactive-p' did" (and still does)

...then just use `interactive-p'.

(I do.  But then, I also want to be backward compatible.)

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

On Wed, Jun 29, 2016 at 5:19 PM, Drew Adams <drew.adams@oracle.com> wrote:

> If (called-interactively-p 'interactive) is equivalent to
> `interactive-p', and if "the call is for a function of no
> arguments that does what `interactive-p' did" (and still does)
>
> ...then just use `interactive-p'.

I would/did but the byte-compiler keeps telling me it is deprecated
and I don't want to see that for this particular function.
Is there a way to turn off such errors for particular functions?

Bob

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

On Wed, Jun 29, 2016 at 5:39 PM, Robert Weiner <rsw@gnu.org> wrote:
> would/did but the byte-compiler keeps telling me it is deprecated
> and I don't want to see that for this particular function.
> Is there a way to turn off such errors for particular functions?

I found the documentation on how to do this:

byte-compile-warnings is a variable defined in ‘bytecomp.el.gz’.
Its value is t

If the list begins with ‘not’, then the remaining elements specify warnings to
suppress.  For example, (not mapcar) will suppress warnings about mapcar.

So I can suppress the warnings locally and have nice looking code but
all installers of my packages will still have to look at the byte
compile warnings then.  I wonder if this could be set within the
*-pkg.el file for a package so that it applies to all users too.

Or I guess I could redefine called-interactively-p to take an optional
parameter inside my packages if John does not want to try that out
given this discussion.

Bob

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

There is also with-no-warnings, that allows you to suppress the warning
at a particular place in the code, not everywhere.

-- 
Dr Richard Stallman
President, Free Software Foundation (gnu.org, fsf.org)
Internet Hall-of-Famer (internethalloffame.org)
Skype: No way! See stallman.org/skype.html.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Clément Pit--Claudel]

[-- Attachment #1.1: Type: text/plain, Size: 451 bytes --]

On 2016-06-29 16:54, Robert Weiner wrote:
> There was a function, interactive-p, that many people used without any
> arguments and found that useful.  Now (called-interactively-p
> 'interactive) is equivalent but makes code much uglier (I know as I
> just changed all my code to use it and it is much less readable now).

Hyperbold is pretty large, so you could also define (hyp-interactive-p) to alias (called-interactively-p 'interactive).


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

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Dmitry Gutov]

On 06/30/2016 12:21 AM, Clément Pit--Claudel wrote:

> Hyperbold is pretty large, so you could also define (hyp-interactive-p) to alias (called-interactively-p 'interactive).

+1.

If called-interactive-p is such a nuisance for Robert, this sounds like 
the best option to me.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

On Wed, Jun 29, 2016 at 5:52 PM, Dmitry Gutov <dgutov@yandex.ru> wrote:
> On 06/30/2016 12:21 AM, Clément Pit--Claudel wrote:
>
>> Hyperbold is pretty large, so you could also define (hyp-interactive-p) to
>> alias (called-interactively-p 'interactive).
>
> If called-interactive-p is such a nuisance for Robert, this sounds like the
> best option to me.

This is not a bad idea but I think suppressing the warnings on
(interactive-p) leaves me with the highest readability for now, so I
am going to pursue that and we'll see how it goes.

Bob

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[raman]

Along those lines I had once suggested that we make (interactive-p) a
synonym for (called-interactively-p 'interactive)  but it got rejected
through a complex email thread. That might still be worth doing.


Also, called-interactively-p as it exists  has other issues --
specifically  fails when called within advice forms. See related email
discussion from about a year ago -- as well as comments in the advice
implementation. --- for the record interactive-p used to work fine in
 advice forms for nearly 18 years --- it broke sometime during the
 transition from advice.el to advice.el plus nadvice.el and the arrival
 of called-interactively-p, all of which (relatively speaking) happened
 at around the same time.
-- 

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

On Wed, Jun 29, 2016 at 10:51 PM, raman <raman@google.com> wrote:
> Along those lines I had once suggested that we make (interactive-p) a
> synonym for (called-interactively-p 'interactive)  but it got rejected
> through a complex email thread. That might still be worth doing.

Maybe the simplest solution would be to undeprecate (interactive-p),
if it is not doing anything wrong, and simply continue in its
documentation to refer to the other calling form for more complex
cases.

I have seen another case where a function has been deprecated
(find-tag, with a suggestion to use xref-find-definitions) where the
new function's calling signature and behavior is totally incompatible
with that of the deprecated function.  This seems wrong.  If the old
behavior works fine and is used widely and you want to add new
behavior, there should be some good thought towards backwards
compatibility, as I am sure is typically done.

It would be nice if John and Eli could just make some final decision
on the interactive-p issue and then put that one to rest, whatever it
is.

Bob

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Eli Zaretskii]

> From: Robert Weiner <rsw@gnu.org>
> Date: Thu, 30 Jun 2016 10:25:26 -0400
> Cc: Richard Stallman <rms@gnu.org>, emacs-devel <emacs-devel@gnu.org>
> 
> It would be nice if John and Eli could just make some final decision
> on the interactive-p issue and then put that one to rest, whatever it
> is.

I have no opinion on the subject, except the general principle that we
shouldn't back up on past decisions unless they are truly disastrous,
because such back-and-forth confuses users and wastes maintenance
efforts.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Andreas Röhler]

On 30.06.2016 17:30, Eli Zaretskii wrote:
>> From: Robert Weiner <rsw@gnu.org>
>> Date: Thu, 30 Jun 2016 10:25:26 -0400
>> Cc: Richard Stallman <rms@gnu.org>, emacs-devel <emacs-devel@gnu.org>
>>
>> It would be nice if John and Eli could just make some final decision
>> on the interactive-p issue and then put that one to rest, whatever it
>> is.
> I have no opinion on the subject, except the general principle that we
> shouldn't back up on past decisions unless they are truly disastrous,

They are. This is another example from a couple of issues which went 
wrong last years.
Code was made swollen up and complicated, simple solutions rejected.

BTW a characteristic design-flaw beyond mentioned last days is around 
use-region-p.
With use-empty-active-region or region-active-p it hardly comes to an end.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Alan Mackenzie]

Hello, Andreas.

On Sun, Jul 03, 2016 at 08:36:56PM +0200, Andreas Röhler wrote:


> On 30.06.2016 17:30, Eli Zaretskii wrote:
> >> From: Robert Weiner <rsw@gnu.org>
> >> Date: Thu, 30 Jun 2016 10:25:26 -0400
> >> Cc: Richard Stallman <rms@gnu.org>, emacs-devel <emacs-devel@gnu.org>

> >> It would be nice if John and Eli could just make some final
> >> decision on the interactive-p issue and then put that one to rest,
> >> whatever it is.
> > I have no opinion on the subject, except the general principle that
> > we shouldn't back up on past decisions unless they are truly
> > disastrous,

> They are. This is another example from a couple of issues which went
> wrong last years.  Code was made swollen up and complicated, simple
> solutions rejected.

This is the sort of thing liable to make people unhappy with your
postings.  Please say exactly what has become turgid and complicated,
where simple solutions have been rejected.  Possibly those simple
solutions were too simple.

> BTW a characteristic design-flaw beyond mentioned last days is around
> use-region-p.  With use-empty-active-region or region-active-p it
> hardly comes to an end.

What hardly comes to an end?  Are you talking about a loop in the code,
our discussion on emacs-devel, or a design flaw not coming to an end?

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Andreas Röhler]

On 03.07.2016 20:56, Alan Mackenzie wrote:
> Hello, Andreas.
>
> On Sun, Jul 03, 2016 at 08:36:56PM +0200, Andreas Röhler wrote:
>
>
>> On 30.06.2016 17:30, Eli Zaretskii wrote:
>>>> From: Robert Weiner <rsw@gnu.org>
>>>> Date: Thu, 30 Jun 2016 10:25:26 -0400
>>>> Cc: Richard Stallman <rms@gnu.org>, emacs-devel <emacs-devel@gnu.org>
>>>> It would be nice if John and Eli could just make some final
>>>> decision on the interactive-p issue and then put that one to rest,
>>>> whatever it is.
>>> I have no opinion on the subject, except the general principle that
>>> we shouldn't back up on past decisions unless they are truly
>>> disastrous,
>> They are. This is another example from a couple of issues which went
>> wrong last years.  Code was made swollen up and complicated, simple
>> solutions rejected.
> This is the sort of thing liable to make people unhappy with your
> postings.

Hmm, that's not the intended thing.

>    Please say exactly what has become turgid and complicated,
> where simple solutions have been rejected.  Possibly those simple
> solutions were too simple.

The moment interactive-p was deprecated against the form discussed here, 
argued detailed against it. IIRC also did Drew. It already took hours - 
in vain so far.


>
>> BTW a characteristic design-flaw beyond mentioned last days is around
>> use-region-p.  With use-empty-active-region or region-active-p it
>> hardly comes to an end.
> What hardly comes to an end?  Are you talking about a loop in the code,
> our discussion on emacs-devel, or a design flaw not coming to an end?
>

The redundancy should be visible by the symbols given.
Why to employ three symbols pointing at the same?

This also was discussed in extenso and a simple form replaced by a 
complicated one.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Eli Zaretskii]

> From: Andreas Röhler <andreas.roehler@online.de>
> Date: Sun, 3 Jul 2016 20:36:56 +0200
> 
> On 30.06.2016 17:30, Eli Zaretskii wrote:
> >> From: Robert Weiner <rsw@gnu.org>
> >> Date: Thu, 30 Jun 2016 10:25:26 -0400
> >> Cc: Richard Stallman <rms@gnu.org>, emacs-devel <emacs-devel@gnu.org>
> >>
> >> It would be nice if John and Eli could just make some final decision
> >> on the interactive-p issue and then put that one to rest, whatever it
> >> is.
> > I have no opinion on the subject, except the general principle that we
> > shouldn't back up on past decisions unless they are truly disastrous,
> 
> They are.

Once again, please back up your assertions with some clear-cut
evidence.  A disastrous decision should not be hard to show as such:
just come up with a couple of examples where this decision causes a
disaster.  This:

> This is another example from a couple of issues which went 
> wrong last years.
> Code was made swollen up and complicated, simple solutions rejected.
> 
> BTW a characteristic design-flaw beyond mentioned last days is around 
> use-region-p.
> With use-empty-active-region or region-active-p it hardly comes to an end.

doesn't even come close.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[raman]

Undeprecating it would be a good solution IMHO, but it was also rejected
at the time, perhaps there has been enough water under the bridge that
one could revisit that.


-- 

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

  > > Telling people to ignore the choice, or choose at random, while
  > > the choice is still necessary, would not be better design.

  > That is an argument for never allowing optional parameters to
  > functions.

Only in certain special cases, where the omission of the argument
is a pitfall for specific reasons.

I am not sure what is right to do here, and I don't want to study the
issue enough to reach a conclusion.  So I don't have views about what
we should do.  I just disagreed with a particular argument.

-- 
Dr Richard Stallman
President, Free Software Foundation (gnu.org, fsf.org)
Internet Hall-of-Famer (internethalloffame.org)
Skype: No way! See stallman.org/skype.html.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[John Wiegley]

[-- Attachment #1: Type: text/plain, Size: 1701 bytes --]

>>>>> Glenn Morris <rgm+news@gnu.org> writes:

> It's easy to search for prior discussion on this. Eg
> http://lists.gnu.org/archive/html/emacs-devel/2015-07/msg00326.html

>    That would defeat the whole purpose (which is to force the author to make
>    a choice between the two kinds, rather than to use one without knowing
>    that there's a choice).

I do not think this is a good argument. Desiring authors to know about such a
distinction is one thing; so document it well. But forcing authors who *do*
know the distinction to be explicit about it always -- just so they don't miss
out on becoming educated the first time -- is poor design.

Simple and common patterns should be simple, because they are common. They
should not be made laborious "to promote education". If a developer is
uneducated about the options, they can be made aware. This isn't the only
place where a mistake can be made by choosing an incorrect option without
warning.

Is a bad argument to `called-interactively-p' really so dangerous that we
require a conscious choice for every point of use? Cargo-culting will result
in this explicit option being copied without understanding anyway, so the
explicitness is not a guarantee against misuse.

I'm completely OK with a patch to undeprecate (interactive-p), even if this
reverses previous decisions. I do not feel beholden to past thinking that
makes little sense to me, just for petty consistency; I trust people can learn
to adapt to any change we make (or, in this case, unmake).

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 629 bytes --]

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Dmitry Gutov]

On 07/06/2016 03:09 AM, John Wiegley wrote:

> I do not think this is a good argument. Desiring authors to know about such a
> distinction is one thing; so document it well. But forcing authors who *do*
> know the distinction to be explicit about it always -- just so they don't miss
> out on becoming educated the first time -- is poor design.

I think the reasons for making it verbose and in-your-face are rather 
clearly decribed in the last paragraphs of called-interactively-p's 
docstring.

As the docstring explains, the function in question should be used 
rarely, if at all. It should follow that the verbosity of its calling 
convention can't be a significant problem.

> Simple and common patterns should be simple, because they are common.

"simple and common", yes.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[John Wiegley]

>>>>> Dmitry Gutov <dgutov@yandex.ru> writes:

> As the docstring explains, the function in question should be used rarely,
> if at all. It should follow that the verbosity of its calling convention
> can't be a significant problem.

If this is the case, I'm curious why this is an issue at all then. Bob, is
really a problem that needs addressing, or is it just for infrequent
convenience? If it's only the latter, let's leave this sleeping dog to lie.

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[raman]

Here's one example of a large package (Emacspeak) and the number of
times the equivalent of interactive-p is used: It would be interesting
to validate the assertion "it's not used often"  by examining other
large packages as well.

Note: ems-interactive-p is what I wrote when interactive-p got
deprecated.

grep ems-interactive-p *.el | wc
   1607    4781   79080
Lines Of Elisp In Emacspeak: 
18:44:33 raman-glaptop2 lisp $ wc *.el
    ...
  79781  256614 2624155 total

-- 

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Dmitry Gutov]

On 07/06/2016 04:46 AM, raman wrote:
> Here's one example of a large package (Emacspeak) and the number of
> times the equivalent of interactive-p is used: It would be interesting
> to validate the assertion "it's not used often"  by examining other
> large packages as well.

The assertion is "it should not be used often", not "it is not used often".

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Eli Zaretskii]

> From: raman <raman@google.com>
> Date: Tue, 05 Jul 2016 18:46:31 -0700
> Cc: emacs-devel@gnu.org
> 
> Here's one example of a large package (Emacspeak) and the number of
> times the equivalent of interactive-p is used: It would be interesting
> to validate the assertion "it's not used often"  by examining other
> large packages as well.
> 
> Note: ems-interactive-p is what I wrote when interactive-p got
> deprecated.
> 
> grep ems-interactive-p *.el | wc
>    1607    4781   79080
> Lines Of Elisp In Emacspeak: 
> 18:44:33 raman-glaptop2 lisp $ wc *.el
>     ...
>   79781  256614 2624155 total

Did this change after Stefan fixed the bug you reported?  Did you try
to see afterwards whether the problems you recall are still there or
solved?

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[raman]

Eli Zaretskii <eliz@gnu.org> writes:

From memory the problem  with called-interactively-p failing when called
from inside an around advice remain, Stefan fixed part of the bug, but I
also remember him saying that he was putting the rest of it on a stack
for some future fix.>> From: raman <raman@google.com>
>> Date: Tue, 05 Jul 2016 18:46:31 -0700
>> Cc: emacs-devel@gnu.org
>> 
>> Here's one example of a large package (Emacspeak) and the number of
>> times the equivalent of interactive-p is used: It would be interesting
>> to validate the assertion "it's not used often"  by examining other
>> large packages as well.
>> 
>> Note: ems-interactive-p is what I wrote when interactive-p got
>> deprecated.
>> 
>> grep ems-interactive-p *.el | wc
>>    1607    4781   79080
>> Lines Of Elisp In Emacspeak: 
>> 18:44:33 raman-glaptop2 lisp $ wc *.el
>>     ...
>>   79781  256614 2624155 total
>
> Did this change after Stefan fixed the bug you reported?  Did you try
> to see afterwards whether the problems you recall are still there or
> solved?
>

-- 

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Eli Zaretskii]

> From: John Wiegley <jwiegley@gmail.com>
> Date: Tue, 05 Jul 2016 17:31:26 -0700
> Cc: emacs-devel@gnu.org
> 
> >>>>> Dmitry Gutov <dgutov@yandex.ru> writes:
> 
> > As the docstring explains, the function in question should be used rarely,
> > if at all. It should follow that the verbosity of its calling convention
> > can't be a significant problem.
> 
> If this is the case, I'm curious why this is an issue at all then. Bob, is
> really a problem that needs addressing, or is it just for infrequent
> convenience? If it's only the latter, let's leave this sleeping dog to lie.

IMO, if we want to make a rational decision that would make sense to
most everybody, we need first to define the problem.  After reading
this and past discussions, I'm still not sure what the problem(s)
is/are.  What's more, I'm not sure anyone else here understands the
actual problem(s) being brought up.  Large portions of the discussions
seem to be based on distant experiences, which I'm not sure is
relevant nowadays, and the problems are described very vaguely.  A
couple of times that an actual problem was reported (e.g., bug#21083),
it was promptly fixed, but I don't see any updated reports after it
was fixed, it seems people are still claiming the same claims
regardless of the fix.

So: what exactly is the problem?  Is it a convenience issue, or are
there some real problems in the recommended replacements for
'interactive-p' (such as 'called-interactively-p')?  Do we have
reproducible recipes for the problems the recommended practices
allegedly have?  If so, would someone please present these recipes
with the current versions of Emacs, preferably as bug reports?

I'd also ask Stefan to please summarize the reasons for deprecating
'interactive-p' and describe its recommended substitutes for the
popular use cases.

Armed with that knowledge, we can hopefully proceed with the subject:
fix the bugs, clarify the documentation, and make the decision whether
the recommended practices are viable or need some changes.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

On Wed, Jul 6, 2016 at 10:48 AM, Eli Zaretskii <eliz@gnu.org> wrote:
> So: what exactly is the problem?  Is it a convenience issue, or are
> there some real problems in the recommended replacements for
> 'interactive-p' (such as 'called-interactively-p')?

I agree with Eli that a good summary, hopefully with examples, of why
one always needs to give an argument to called-interactively-p would
be helpful and how making its argument optional could cause problems,
if that is indeed the case.

I do think called-interactively-p is a better name for the function
than interactive-p and that is why I originally suggested just making
the argument to it optional, defaulting to the same behavior as when
it is given an 'interactive argument.  My first post in this thread is
a good summary of why the change was suggested, which was for ease of
code writing and reading, not any operational problem.

Bob

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Noam Postavsky]

On Tue, Jul 5, 2016 at 8:09 PM, John Wiegley <jwiegley@gmail.com> wrote:
> Simple and common patterns should be simple, because they are common.

The simplest way to do this if you don't care about the
distinction between (called-interactively 'any)
and (called-interactively 'interactive) is to add an argument
that is set to non-nil by the interactive spec, as described in
the manual:

    The recommended way to test whether the function was called using
    ‘call-interactively’ is to give it an optional argument
    ‘print-message’ and use the ‘interactive’ spec to make it
    non-‘nil’ in interactive calls.  Here’s an example:

     (defun foo (&optional print-message)
       (interactive "p")
       (when print-message
         (message "foo")))

    We use ‘"p"’ because the numeric prefix argument is never ‘nil’.
    Defined in this way, the function does display the message when
    called from a keyboard macro.

       The above method with the additional argument is usually best,
    because it allows callers to say “treat this call as interactive”.

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Robert Weiner]

On Tue, Jul 5, 2016 at 8:24 PM, Noam Postavsky
<npostavs@users.sourceforge.net> wrote:
>     The recommended way to test whether the function was called using
>     ‘call-interactively’ is to give it an optional argument
>     ‘print-message’ and use the ‘interactive’ spec to make it
>     non-‘nil’ in interactive calls.  Here’s an example:
>
>      (defun foo (&optional print-message)
>        (interactive "p")
>        (when print-message
>          (message "foo")))

This makes little sense to me.  Every time I have a function that
needs to test whether it was called interactively you want me to add
essentially an interactive-flag to its argument signature and then
never send this argument when called non-interactively since the
caller won't care and the argument is optional so it will be nil by
default.  Again, why should we name and have to look at an argument to
the function when the interactive test we want is internal to it.
This is similar to the argument that we shouldn't have to send the
default 'interactive argument to called-interactively-p.

I'm with John, let's simplify this.  Let's undeprecate
(interactive-p).  Did people actually have problems with that that
kept coming up?  Raman and I seemed to have used it for years without
encountering any.

And John, if you undeprecate it do you really need a patch or can you
just change it?

Bob

Re: Emacs 25.0.04: Feature Request: Make called-interactively-p's argument optional

[Andreas Röhler]

On 06.07.2016 07:03, Robert Weiner wrote:
> On Tue, Jul 5, 2016 at 8:24 PM, Noam Postavsky
> <npostavs@users.sourceforge.net> wrote:
>>      The recommended way to test whether the function was called using
>>      ‘call-interactively’ is to give it an optional argument
>>      ‘print-message’ and use the ‘interactive’ spec to make it
>>      non-‘nil’ in interactive calls.  Here’s an example:
>>
>>       (defun foo (&optional print-message)
>>         (interactive "p")
>>         (when print-message
>>           (message "foo")))
> This makes little sense to me.  Every time I have a function that
> needs to test whether it was called interactively you want me to add
> essentially an interactive-flag to its argument signature and then
> never send this argument when called non-interactively since the
> caller won't care and the argument is optional so it will be nil by
> default.  Again, why should we name and have to look at an argument to
> the function when the interactive test we want is internal to it.
> This is similar to the argument that we shouldn't have to send the
> default 'interactive argument to called-interactively-p.
>
> I'm with John, let's simplify this.  Let's undeprecate
> (interactive-p).  Did people actually have problems with that that
> kept coming up?  Raman and I seemed to have used it for years without
> encountering any.
>
> And John, if you undeprecate it do you really need a patch or can you
> just change it?
>
> Bob
>

1+