defcustom: order matters in `choice' with value-to-internal

defcustom: order matters in `choice' with value-to-internal

[Reiner Steib]

Hi,

consider these two defcustoms, which only differ in the order of the
choice elements:

--8<---------------cut here---------------start------------->8---
(defcustom rs-test-1
  "^From:\\|^Newsgroups:\\|^Subject:"
  "doc string 1, list of regexps, regexp or nil"
  :type '(choice
	  (const nil :tag "Use gnus-ignored-headers")
	  regexp
	  (repeat :value-to-internal (lambda (widget value)
				       (custom-split-regexp-maybe value))
		  :match (lambda (widget value)
			   (or (stringp value)
			       (widget-editable-list-match widget value)))
		  regexp)))

(defcustom rs-test-2
  "^From:\\|^Newsgroups:\\|^Subject:"
  "doc string 2, list of regexps, regexp or nil"
  :type '(choice
	  (repeat :value-to-internal (lambda (widget value)
				       (custom-split-regexp-maybe value))
		  :match (lambda (widget value)
			   (or (stringp value)
			       (widget-editable-list-match widget value)))
		  regexp)
	  (const nil :tag "Use gnus-ignored-headers")
	  regexp))
--8<---------------cut here---------------end--------------->8---

In the customize buffer, I get (Emacs 22 branch):

,----[ M-x customize-apropos RET rs-test RET ]
| rs-test-1: [Hide Value] [Value Menu] Regexp: ^From:\|^Newsgroups:\|^Subject:
|    [State]: STANDARD.
| 
| doc string 1, list of regexps, regexp or nil
| Groups: [message-faces]
| 
| rs-test-2: [Hide Value] [Value Menu] Repeat:
| [INS] [DEL] Regexp: ^From:
| [INS] [DEL] Regexp: ^Newsgroups:
| [INS] [DEL] Regexp: ^Subject:
| [INS]
|    [State]: STANDARD.
| 
| doc string 2, list of regexps, regexp or nil
| Groups: [message-faces]
`----

I would have expected the same, split result for both variables.

(BTW, I have now idea why "Groups: [message-faces]" is shown.  Seems
like some random group?)

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: defcustom: order matters in `choice' with value-to-internal

[Andreas Schwab]

Reiner Steib <reinersteib+gmane@imap.cc> writes:

> consider these two defcustoms, which only differ in the order of the
> choice elements:
>
> (defcustom rs-test-1
>   "^From:\\|^Newsgroups:\\|^Subject:"
>   "doc string 1, list of regexps, regexp or nil"
>   :type '(choice
> 	  (const nil :tag "Use gnus-ignored-headers")
> 	  regexp
> 	  (repeat :value-to-internal (lambda (widget value)
> 				       (custom-split-regexp-maybe value))
> 		  :match (lambda (widget value)
> 			   (or (stringp value)
> 			       (widget-editable-list-match widget value)))
> 		  regexp)))
>
> (defcustom rs-test-2
>   "^From:\\|^Newsgroups:\\|^Subject:"
>   "doc string 2, list of regexps, regexp or nil"
>   :type '(choice
> 	  (repeat :value-to-internal (lambda (widget value)
> 				       (custom-split-regexp-maybe value))
> 		  :match (lambda (widget value)
> 			   (or (stringp value)
> 			       (widget-editable-list-match widget value)))
> 		  regexp)
> 	  (const nil :tag "Use gnus-ignored-headers")
> 	  regexp))

Order generally matters, the first matching choice is used.  There is
nothing else that favors the repeat over regexp.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: defcustom: order matters in `choice' with value-to-internal

[Reiner Steib]

On Fri, Mar 21 2008, Andreas Schwab wrote:

> Order generally matters, the first matching choice is used.  There is
> nothing else that favors the repeat over regexp.

D'oh, of course.  I must have been confused.  Thanks.

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: defcustom: order matters in `choice' with value-to-internal

[Stefan Monnier]

> Order generally matters, the first matching choice is used.  There is
> nothing else that favors the repeat over regexp.

But AFAICT the only difference between the two is the order between
(const nil ..) and (repeat ...): the ordering between `repeat' and
`regexp' is the same in both cases, so it still looks odd.


        Stefan

Re: defcustom: order matters in `choice' with value-to-internal

[Reiner Steib]

On Sat, Mar 22 2008, Stefan Monnier wrote:

>> Order generally matters, the first matching choice is used.  There is
>> nothing else that favors the repeat over regexp.
>
> But AFAICT the only difference between the two is the order between
> (const nil ..) and (repeat ...): the ordering between `repeat' and
> `regexp' is the same in both cases, so it still looks odd.

No, the order of `repeat regexp' and `regexp' is different:

(defcustom rs-test-1
[...]
  :type '(choice
	  (const nil)
	  regexp
	  (repeat [...] regexp)))

(defcustom rs-test-2
[...]
  :type '(choice
	  (repeat [...] regexp)
	  (const nil)
	  regexp))

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: defcustom: order matters in `choice' with value-to-internal

[Stefan Monnier]

>>> Order generally matters, the first matching choice is used.  There is
>>> nothing else that favors the repeat over regexp.
>> 
>> But AFAICT the only difference between the two is the order between
>> (const nil ..) and (repeat ...): the ordering between `repeat' and
>> `regexp' is the same in both cases, so it still looks odd.

> No, the order of `repeat regexp' and `regexp' is different:

Duh!  Sorry!  Victim of my auto-unwrap hack, which hid the `regexp'
entry at the end of the (const ...) lines


        Stefan