On 4.10.2013, at 20:47, Paul Rudin <paul@rudin.co.uk> wrote:

> Nicolas Goaziou <n.goaziou@gmail.com> writes:
> 
>> Hello,
>> 
>> Paul Rudin <paul@rudin.co.uk> writes:
>> 
>>> Paul Rudin <paul@rudin.co.uk> writes:
>>> 
>>> (about lists)
>>> 
>>> Here is a test for this problem. Notice that if you change the "a)" to "1."
>>> - for example - the test passes. So this is something specific to
>>> alphabetically labelled list items.
>>> 
>>> 
>>> (ert-deftest org-list-item-test ()
>>>  (with-temp-buffer
>>>    (org-mode)
>>>    (let ((org-allow-alphabetical t)
>>> 	  (fill-column 70))
>>>      (insert "1. some stuff\n"
>>> 	      "   a) an alphabetic list item with text longer that the current fill column so that it gets wrapped by fill-paragraph")
>>>      (fill-paragraph)t
>>>      (should (not (equal (org-in-item-p) 1))))))
>> 
>> Quoting `org-list-allow-alphabetical' docstring:
>> 
>>  This variable needs to be set before org.el is loaded.  If you
>>  need to make a change while Emacs is running, use the customize
>>  interface or run the following code, where VALUE stands for the
>>  new value of the variable, after updating it:
>> 
>>    (when (featurep 'org-element) (load "org-element" t t))
>> 
>> So let-binding it cannot work.
> 
> 
> Aha. So how do you test things like that? This seems to be OK. Is it?
> 
> (ert-deftest org-list-item-test ()
>  (with-temp-buffer
>    (org-mode)
>    (let ((fill-column 70))
>      (custom-set-variables '(org-list-allow-alphabetical t))
>      (insert "1. some stuff\n"
> 	      "   a) an alphabetic list item with text longer that the current fill column so that it gets wrapped by fill-paragraph")
>      (fill-paragraph)
>      (should (not (equal (org-in-item-p) 1))))))
> 
> 
> This also explains what I was seeing, as I was setting it in my init
> file after org was loaded. I was lulled into a false sense of security
> because some aspects of alphabetical list label work in this
> circumstance.
> 
> All I need now is roman numeral list labels too :)

No chance!  :)

- Carsten