bug#37916: GNUS: wrong item selected in gnus-thread-hide-subtree customization

bug#37916: GNUS: wrong item selected in gnus-thread-hide-subtree customization

[Sergey Organov]

The "Non-nil" defcustom item is selected in customization buffer when
actual value of the gnus-thread-hide-subtree is 'nil.

-- 8< --

Fix gnus-thread-hide-subtree defcustom

* lisp/gnus/gnus-sum.el (gnus-thread-hide-subtree): fix order of
items to prevent first 'sexp' from matching 'nil, that caused
"Non-nil" item to be selected in customization buffer when actual
variable value is 'nil.
---
 lisp/gnus/gnus-sum.el | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lisp/gnus/gnus-sum.el b/lisp/gnus/gnus-sum.el
index b5d7448..1970465 100644
--- a/lisp/gnus/gnus-sum.el
+++ b/lisp/gnus/gnus-sum.el
@@ -332,11 +332,11 @@ This can be a predicate specifier which says which threads to hide.
 If threads are hidden, you have to run the command
 `gnus-summary-show-thread' by hand or select an article."
   :group 'gnus-thread
-  :type '(radio (sexp :format "Non-nil\n"
+  :type '(radio (const nil)
+		(sexp :format "Non-nil\n"
 		      :match (lambda (widget value)
 			       (not (or (consp value) (functionp value))))
 		      :value t)
-		(const nil)
 		(sexp :tag "Predicate specifier")))
 
 (defcustom gnus-thread-hide-killed t
-- 
2.1.4

bug#37916: GNUS: wrong item selected in gnus-thread-hide-subtree customization

[Lars Ingebrigtsen]

Sergey Organov <sorganov@gmail.com> writes:

> The "Non-nil" defcustom item is selected in customization buffer when
> actual value of the gnus-thread-hide-subtree is 'nil.

Hm...  'nil and nil are the same.

[...]

> -  :type '(radio (sexp :format "Non-nil\n"
> +  :type '(radio (const nil)
> +		(sexp :format "Non-nil\n"
>  		      :match (lambda (widget value)
>  			       (not (or (consp value) (functionp value))))
>  		      :value t)
> -		(const nil)

Oh, I see.  I think the problem is that that :match is just wrong.
Changing the order of the values (as this patch does) also makes the
problem go away, but I think the right fix here is to fix the :match.

So I've installed the following patch instead:

diff --git a/lisp/gnus/gnus-sum.el b/lisp/gnus/gnus-sum.el
index b5d744843f..f21bc7584e 100644
--- a/lisp/gnus/gnus-sum.el
+++ b/lisp/gnus/gnus-sum.el
@@ -334,7 +334,7 @@ gnus-thread-hide-subtree
   :group 'gnus-thread
   :type '(radio (sexp :format "Non-nil\n"
 		      :match (lambda (widget value)
-			       (not (or (consp value) (functionp value))))
+			       (and value (not (functionp value))))
 		      :value t)
 		(const nil)
 		(sexp :tag "Predicate specifier")))


-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#37916: GNUS: wrong item selected in gnus-thread-hide-subtree customization

[Sergey Organov]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Sergey Organov <sorganov@gmail.com> writes:
>
>> The "Non-nil" defcustom item is selected in customization buffer when
>> actual value of the gnus-thread-hide-subtree is 'nil.
>
> Hm...  'nil and nil are the same.
>
> [...]
>
>> -  :type '(radio (sexp :format "Non-nil\n"
>> +  :type '(radio (const nil)
>> +		(sexp :format "Non-nil\n"
>>  		      :match (lambda (widget value)
>>  			       (not (or (consp value) (functionp value))))
>>  		      :value t)
>> -		(const nil)
>
> Oh, I see.  I think the problem is that that :match is just wrong.
> Changing the order of the values (as this patch does) also makes the
> problem go away, but I think the right fix here is to fix the :match.

Yes, I also thought it is, but took kludgy way as I'm not familiar
with the :math specifications.

Thanks for fixing it the right way!

-- Sergey