Unclear org-icompleting-read interface

Unclear org-icompleting-read interface

[Oleh Krehel]

Hi all,

The current `org-icompleting-read' is defined as taking only `&rest
args', which can be as many as 8, all of them without a name.

I propose to update the definition to:

    (cl-defun org-icompleting-read (prompt collection
                                    &key predicate require-match
                                      initial-input hist def inherit-input-method)
      ;; ...
      )

Let me know if you like the idea. If so, there are 43 calls to
`org-icompleting-read' in the source that need to be updated. Some of
them should as they are, since usually only the first two arguments are
used.

This change would make the calls to `org-icompleting-read' much easier
to read.

regards,
Oleh

Re: Unclear org-icompleting-read interface

[Nicolas Goaziou]

Hello,

Oleh Krehel <ohwoeowho@gmail.com> writes:

> The current `org-icompleting-read' is defined as taking only `&rest
> args', which can be as many as 8, all of them without a name.
>
> I propose to update the definition to:
>
>     (cl-defun org-icompleting-read (prompt collection
>                                     &key predicate require-match
>                                       initial-input hist def inherit-input-method)
>       ;; ...
>       )

I'd rather not use cl-defun in code base, which is slightly more
difficult to understand, and hardly unavoidable. Moreover,
`org-icompleting-read' is not harder to read as `completing-read'
anyway.

> This change would make the calls to `org-icompleting-read' much easier
> to read.

While we're at it, I suggest instead to make `org-icompleting-read' an
obsolete alias for `completing-read' in Org 8.4.
`partial-completing-mode' is obsolete since Emacs 24.1.


Regards,

-- 
Nicolas Goaziou

Re: Unclear org-icompleting-read interface

[Oleh Krehel]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Moreover,
> `org-icompleting-read' is not harder to read as `completing-read'
> anyway.

I disagree on this. `completing-read' at least declares its own argument
list. So it's possible to distinguish which argument is which, albeit
with some effort for a large amount of arguments.

For example, in this case there are 5 args, the first two I know by
heart. Not the other three.

    (setq prop (org-icompleting-read
                "Property: " (mapcar 'list (org-buffer-property-keys t nil t))
                nil nil prop))

And the arglist: "&rest args", instead of "prompt collection predicate
require-match initial-input hist def inherit-input-method".

Here's how it would look like for a `cl-defun' variant:

    (org-icompleting-read
     "Property: " (mapcar 'list (org-buffer-property-keys t nil t))
     :initial-input prop)

Down to 3 arguments from 5, and the last one is very easy to understand,
even without eldoc.

In any case this also solves the problem of the argument list:

    (define-obsolete-function-alias
        'org-icompleting-read 'completing-read)

Re: Unclear org-icompleting-read interface

[Rasmus]

Hi,

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Oleh Krehel <ohwoeowho@gmail.com> writes:
>
>> The current `org-icompleting-read' is defined as taking only `&rest
>> args', which can be as many as 8, all of them without a name.
>>
>> I propose to update the definition to:
>>
>>     (cl-defun org-icompleting-read (prompt collection
>>                                     &key predicate require-match
>>                                       initial-input hist def inherit-input-method)
>>       ;; ...
>>       )
>
> I'd rather not use cl-defun in code base, which is slightly more
> difficult to understand, and hardly unavoidable. Moreover,
> `org-icompleting-read' is not harder to read as `completing-read'
> anyway.

I agree.  It's confusing how the docstring and the λ is written, but it's
just a "front-end" for completing-read.  A reference to completing-read
should be added to the docstring.

>> This change would make the calls to `org-icompleting-read' much easier
>> to read.
>
> While we're at it, I suggest instead to make `org-icompleting-read' an
> obsolete alias for `completing-read' in Org 8.4.
> `partial-completing-mode' is obsolete since Emacs 24.1.

That was my "feeling" about org-icompleting-read as well last time I
looked at it (also a message by Oleh).  For org 8.4 we can also get rid of
all the iswitchb stuff.  Good times ahead...

Rasmus

-- 
This is the kind of tedious nonsense up with which I will not put

Re: Unclear org-icompleting-read interface

[Nicolas Goaziou]

Hello,

Rasmus <rasmus@gmx.us> writes:

> I agree.  It's confusing how the docstring and the λ is written, but it's
> just a "front-end" for completing-read.  A reference to completing-read
> should be added to the docstring.

Isn't it already the case?

> That was my "feeling" about org-icompleting-read as well last time I
> looked at it (also a message by Oleh).  For org 8.4 we can also get rid of
> all the iswitchb stuff.  Good times ahead...

Indeed. Can't wait...

Regards,

-- 
Nicolas Goaziou

Re: Unclear org-icompleting-read interface

[Rasmus]

Hi,

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Rasmus <rasmus@gmx.us> writes:
>
>> I agree.  It's confusing how the docstring and the λ is written, but it's
>> just a "front-end" for completing-read.  A reference to completing-read
>> should be added to the docstring.
>
> Isn't it already the case?

I don't know.  I have been saying stupid things recently.  Lack of sleep
is probably the reason, but I like to think that somehow my org version is
messed up.


     org-icompleting-read is a compiled Lisp function in /usr/share/emacs/site-lisp/org/

     (org-icompleting-read &rest ARGS)

     Completing-read using ‘ido-mode’ or ‘iswitchb’ speedups if available.

A lot of older org functions have terrible documentation.  I had to study
the code, several calls of org-icompleting-read, and completion-read to
understand what it was doing.

>> That was my "feeling" about org-icompleting-read as well last time I
>> looked at it (also a message by Oleh).  For org 8.4 we can also get rid of
>> all the iswitchb stuff.  Good times ahead...
>
> Indeed. Can't wait...

Me to.  I hope I will have more time in Eternal September.

Rasmus

-- 
May the Force be with you