inverse of add-to-list: remove-from-list

inverse of add-to-list: remove-from-list

[Uwe Brauer]

Hi

This has always bothered me: for example 

(add-to-list 'org-export-filter-src-block-functions
         'my-html-filter-src-blocks)

Adds a function to the variable org-export-filter-src-block-function

But there is, as far as I know, no inverse functionality 

(remove-from-list 'org-export-filter-src-block-functions
         'my-html-filter-src-blocks)

Does not exist. Wouldn't that be useful?

Reason there is add-hook and remove-hook

Regards

Uwe Brauer 

Re: inverse of add-to-list: remove-from-list

[Andreas Schwab]

On Okt 13 2020, Uwe Brauer wrote:

> But there is, as far as I know, no inverse functionality 
>
> (remove-from-list 'org-export-filter-src-block-functions
>          'my-html-filter-src-blocks)
>
> Does not exist. Wouldn't that be useful?

Doesn't remove do what you need?

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: inverse of add-to-list: remove-from-list

[Thibaut Verron]

Le mar. 13 oct. 2020 à 11:10, Andreas Schwab <schwab@linux-m68k.org> a écrit :
>
> On Okt 13 2020, Uwe Brauer wrote:
>
> > But there is, as far as I know, no inverse functionality
> >
> > (remove-from-list 'org-export-filter-src-block-functions
> >          'my-html-filter-src-blocks)
> >
> > Does not exist. Wouldn't that be useful?
>
> Doesn't remove do what you need?

Note that the syntax is slightly different: the element comes second,
and the list is unquoted.

Re: inverse of add-to-list: remove-from-list

[Uwe Brauer]

[-- Attachment #1: Type: text/plain, Size: 954 bytes --]

>>> "TV" == Thibaut Verron <thibaut.verron@gmail.com> writes:

> Le mar. 13 oct. 2020 à 11:10, Andreas Schwab <schwab@linux-m68k.org> a écrit :
>> 
>> On Okt 13 2020, Uwe Brauer wrote:
>> 
>> > But there is, as far as I know, no inverse functionality
>> >
>> > (remove-from-list 'org-export-filter-src-block-functions
>> >          'my-html-filter-src-blocks)
>> >
>> > Does not exist. Wouldn't that be useful?
>> 
>> Doesn't remove do what you need?

> Note that the syntax is slightly different: the element comes second,
> and the list is unquoted.

Do you mean 

(setq org-export-filter-src-block-functions (remove 'my-html-filter-src-blocks org-export-filter-src-block-functions))

While:

(add-to-list 'org-export-filter-src-block-functions
         'my-html-filter-src-blocks)


I think 

(remove-from-list 'org-export-filter-src-block-functions
         'my-html-filter-src-blocks)

Would be easier to recall.

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5673 bytes --]

Re: inverse of add-to-list: remove-from-list

[Stefan Monnier]

Uwe Brauer [2020-10-13 10:42:46] wrote:
> This has always bothered me: for example 
>
> (add-to-list 'org-export-filter-src-block-functions
>          'my-html-filter-src-blocks)

I think this reflects a misdesign in `org-export-filters-alist`.
It should use hooks rather than lists of functions.  This way you'd use
`add-hook` instead of `add-to-list`.

> (remove-from-list 'org-export-filter-src-block-functions
>          'my-html-filter-src-blocks)
>
> Does not exist. Wouldn't that be useful?

I think if we want to add something to remove an element from a variable
holding a list, then we should make it accept a "place" rather than
a symbol as argument.  I.e. make it the opposite of `push` (or
`cl-pushnew`) rather than the opposite of `add-to-list`.


        Stefan "who dislikes `add-to-list`"

Re: inverse of add-to-list: remove-from-list

[Boruch Baum]

> On Okt 13 2020, Uwe Brauer wrote:
> > But there is, as far as I know, no inverse functionality

1] Similar to what Andreas suggested, try this:

     (let ((aa '(1 2 3 1 4 1 2)))
       (delete 2 aa)
       aa)

   And note that the fine distinction made in the docstring for the
   `delete' function:

     "If SEQ is not a list, deletion is never performed destructively;
      instead this function creates and returns a new vector or string.

      Write ‘(setq foo (delete element foo))’ to be sure of correctly
      changing the value of a sequence ‘foo’.

   So, in the case of SEQ being a list, there is no need to `setq foo'.

2] That really could/should be made clearer in the docstring; something
   like:

  "If SEQ is a list, the original symbol is modified in place, ie.
   destructively. If SEQ is not a list, eg. a vector or string, this
   function creates and returns a new vector or string, and you must
   explicitly ‘(setq foo (delete element foo))’ to alter the original."

  If there's agreement on this docstring change, can it be committed?

3] Emacs also offers a more flexible function `cl-delete'. See there.

--
hkp://keys.gnupg.net
CA45 09B5 5351 7C11 A9D1  7286 0036 9E45 1595 8BC0

Re: inverse of add-to-list: remove-from-list

[Andreas Schwab]

On Okt 13 2020, Boruch Baum wrote:

>> On Okt 13 2020, Uwe Brauer wrote:
>> > But there is, as far as I know, no inverse functionality
>
> 1] Similar to what Andreas suggested, try this:
>
>      (let ((aa '(1 2 3 1 4 1 2)))
>        (delete 2 aa)
>        aa)
>
>    And note that the fine distinction made in the docstring for the
>    `delete' function:
>
>      "If SEQ is not a list, deletion is never performed destructively;
>       instead this function creates and returns a new vector or string.
>
>       Write ‘(setq foo (delete element foo))’ to be sure of correctly
>       changing the value of a sequence ‘foo’.
>
>    So, in the case of SEQ being a list, there is no need to `setq foo'.

This is not true.  If the element to delete is the first element of the
list, it returns the cdr of the list.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: inverse of add-to-list: remove-from-list

[Boruch Baum]

On 2020-10-13 22:01, Andreas Schwab wrote:
> On Okt 13 2020, Boruch Baum wrote:
>
> >> On Okt 13 2020, Uwe Brauer wrote:
> >> > But there is, as far as I know, no inverse functionality
> >
> > 1] Similar to what Andreas suggested, try this:
> >
> >      (let ((aa '(1 2 3 1 4 1 2)))
> >        (delete 2 aa)
> >        aa)
> >
> >    And note that the fine distinction made in the docstring for the
> >    `delete' function:
> >
> >      "If SEQ is not a list, deletion is never performed destructively;
> >       instead this function creates and returns a new vector or string.
> >
> >       Write ‘(setq foo (delete element foo))’ to be sure of correctly
> >       changing the value of a sequence ‘foo’.
> >
> >    So, in the case of SEQ being a list, there is no need to `setq foo'.
>
> This is not true.

Oops. You're correct...

>  If the element to delete is the first element of the list, it returns
> the cdr of the list.

... but to quibble, that's not exactly what I'm getting

  (let ((aa '(1 2 3 1 4 1 2)))
    (delete 1 aa)
    aa)

returns (1 2 3 4 2), so it seems to destructively modify the cdr of the
list.

--
hkp://keys.gnupg.net
CA45 09B5 5351 7C11 A9D1  7286 0036 9E45 1595 8BC0

Re: inverse of add-to-list: remove-from-list

[Andreas Schwab]

On Okt 13 2020, Boruch Baum wrote:

> ... but to quibble, that's not exactly what I'm getting
>
>   (let ((aa '(1 2 3 1 4 1 2)))
>     (delete 1 aa)
>     aa)
>
> returns (1 2 3 4 2),

QED.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: inverse of add-to-list: remove-from-list

[Teemu Likonen]

[-- Attachment #1: Type: text/plain, Size: 2214 bytes --]

* 2020-10-13 15:34:00-04, Boruch Baum wrote:

>      (let ((aa '(1 2 3 1 4 1 2)))
>        (delete 2 aa)
>        aa)

In Lisp world it is not recommended to use destructive operations like
DELETE for literal objects like list created with QUOTE operator '(1 2)
or a string created with quotation marks ("abc"). Compilers can create a
single object for all literally defined object. Emacs Lisp manual
"(elisp) Equality Predicates" says:

    The Emacs Lisp byte compiler may collapse identical literal objects,
    such as literal strings, into references to the same object, with
    the effect that the byte-compiled code will compare such objects as
    ‘eq’, while the interpreted version of the same code will not.

For example:

    (defvar foo '(1 2 3))

    (let ((bar '(1 2 3)))
      (eq foo bar)               ; This may return T.
      (setq bar (delete 3 bar))) ; This may also modify FOO.

So, using a destructive operation for a literal object may change it
everywhere. To create a modifiable objects we should use LIST,
MAKE-LIST, MAKE-STRING, COPY-SEQUENCE, CL-COPY-SEQ, for example.

>    And note that the fine distinction made in the docstring for the
>    `delete' function:
>
>      "If SEQ is not a list, deletion is never performed destructively;
>       instead this function creates and returns a new vector or string.
>
>       Write ‘(setq foo (delete element foo))’ to be sure of correctly
>       changing the value of a sequence ‘foo’.
>
>    So, in the case of SEQ being a list, there is no need to `setq foo'.

In Lisp manuals "destructive" means that the operation may modify the
original object given as an argument. It doesn't promise anything else.
Sometimes the original object is modified, sometimes it is not. Even if
the original object is modified it doesn't promise that it's the same as
the return value of a destructive function. The original object and the
return value can be partially the same: lists may share some cons cells.
We must always use the return value of a destructive function.

-- 
/// Teemu Likonen - .-.. https://www.iki.fi/tlikonen/
// OpenPGP: 4E1055DC84E9DFF613D78557719D69D324539450

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 251 bytes --]

Re: inverse of add-to-list: remove-from-list

[Thibaut Verron]

Le mar. 13 oct. 2020 à 15:58, Stefan Monnier
<monnier@iro.umontreal.ca> a écrit :
>
> Uwe Brauer [2020-10-13 10:42:46] wrote:
> > This has always bothered me: for example
> >
> > (add-to-list 'org-export-filter-src-block-functions
> >          'my-html-filter-src-blocks)
>
> I think this reflects a misdesign in `org-export-filters-alist`.
> It should use hooks rather than lists of functions.  This way you'd use
> `add-hook` instead of `add-to-list`.

I might be missing something, but where would the harm be in using

> (add-hook 'org-export-filter-src-block-functions 'my-html-filter-src-blocks)

even if it is not a hook?

> I think if we want to add something to remove an element from a variable
> holding a list, then we should make it accept a "place" rather than
> a symbol as argument.  I.e. make it the opposite of `push` (or
> `cl-pushnew`) rather than the opposite of `add-to-list`.
>
> Stefan "who dislikes `add-to-list`"

It looks like for that we already have remove and cl-remove indeed.

So basically the answer would be, instead of looking for an opposite
to add-to-list, use cl-pushnew which has a natural opposite in
cl-remove. Is that so?

For an exact opposite to remove, should there be an alias to the wrapper below?

> (lambda (elt place) (pushnew elt place :test 'equal))

Re: inverse of add-to-list: remove-from-list

[Uwe Brauer]

[-- Attachment #1: Type: text/plain, Size: 277 bytes --]

>>> "TV" == Thibaut Verron <thibaut.verron@gmail.com> writes:

> I might be missing something, but where would the harm be in using

> (add-hook 'org-export-filter-src-block-functions 'my-html-filter-src-blocks)

> even if it is not a hook?

I'd love to clarify that question!

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5673 bytes --]

Re: inverse of add-to-list: remove-from-list

[Stefan Monnier]

> I might be missing something, but where would the harm be in using
>> (add-hook 'org-export-filter-src-block-functions 'my-html-filter-src-blocks)
> even if it is not a hook?

I think none, but it would be harmful to do:

    (add-hook 'org-export-filter-src-block-functions
              #'my-html-filter-src-blocks
              nil t)

> So basically the answer would be, instead of looking for an opposite
> to add-to-list, use cl-pushnew which has a natural opposite in
> cl-remove. Is that so?

`cl-remove` is not the opposite of `cl-pushnew`, because

    (cl-pushnew V X :test F)
    (cl-remove  V X :test F)

will usually result in X being different from what it started (because
`cl-remove` does not modify its argument X).


        Stefan

Re: inverse of add-to-list: remove-from-list

[Thibaut Verron]

Le jeu. 15 oct. 2020 à 18:31, Stefan Monnier
<monnier@iro.umontreal.ca> a écrit :
>
> > I might be missing something, but where would the harm be in using
> >> (add-hook 'org-export-filter-src-block-functions 'my-html-filter-src-blocks)
> > even if it is not a hook?
>
> I think none, but it would be harmful to do:
>
>     (add-hook 'org-export-filter-src-block-functions
>               #'my-html-filter-src-blocks
>               nil t)

Oh, right, thanks.

> > So basically the answer would be, instead of looking for an opposite
> > to add-to-list, use cl-pushnew which has a natural opposite in
> > cl-remove. Is that so?
>
> `cl-remove` is not the opposite of `cl-pushnew`, because
>
>     (cl-pushnew V X :test F)
>     (cl-remove  V X :test F)
>
> will usually result in X being different from what it started (because
> `cl-remove` does not modify its argument X).

Oops. I guess it goes to say how confusing the situation is. :)