Re: [Patch] to correctly sort the items with emphasis marks in a list

[Juan Manuel Macías]

Hello again,

Since I have an old Emacs version (24.5.1) on my Raspberry, I've done
a few more tests. The situation is the following:

1. On Arch Linux and Emacs 27.2:

- The system locales are set to "es_ES.UTF-8". When, inside Emacs, I do M-x
getenv RET LANG RET, I get "es_ES.UTF-8".

org-sort-list a -> wrong result;

- Launching Emacs from terminal with

LANG=es_ES.UTF-8 \
etc...

org-sort-list a -> wrong result again.

2. On Fedora 32 (virtual machine) and Emacs 27.1

- Everything as in the previous case.

3. On Raspian stretch and Emacs 24.5.1:

- The system locales are set to "es_ES.UTF-8" as well. When, inside a
normal Emacs session, I do M-x getenv RET lang RET, I get "es_ES.UTF-8".

org-sort-list a -> wrong result;

- Launching Emacs from terminal with

LANG=es_ES.UTF-8 \
etc...

In this case the list is ordered correctly, but I observe that similar
forms with or without emphasis are not always ordered in the same way.
Sometimes the non-emphasized forms are ordered before and sometimes
they are ordered after (?).

I don't know if I'm missing something...

Best regards,

Juan Manuel

Juan Manuel Macías writes:

> Hi Maxim,
>
> Thanks again. In my case, I keep getting the wrong result. In addition
> to the test I did in a virtual machine with Fedora, I use Arch Linux in
> my every day computers, with locales correctly (I hope) configured as
> es_ES.UTF-8 (there was a typo in my previous mail, sorry:
> 'en_ES.UTF-8'). In my /etc/locale.conf file I have:
>
> LANG=es_ES.UTF-8
> LC_ADDRESS=es_ES.UTF-8
> LC_IDENTIFICATION=es_ES.UTF-8
> LC_MEASUREMENT=es_ES.UTF-8
> LC_MONETARY=es_ES.UTF-8
> LC_NAME=es_ES.UTF-8
> LC_NUMERIC=es_ES.UTF-8
> LC_PAPER=es_ES.UTF-8
> LC_TELEPHONE=es_ES.UTF-8
> LC_TIME=es_ES.UTF-8
>
> And with locale -a, I get a list of enabled locales:
>
> C
> en_US.utf8
> es_ES.utf8
> POSIX
>
> However, I keep getting the wrong result :-( (Arch, Emacs 27.2).
>
> Even with
>
> LANG=en_EN.UTF-8 \
> emacs -nw -Q -L ~/src/org-mode/lisp/ -L ~/src/org-mode/contrib/lisp/ \
> list.org
>
> Maybe the problem is on Arch's side (?), although it was also reproducible
> in the test I did with Fedora in virtual machine and Emacs 27.
>
> Best regards,
>
> Juan Manuel
>
> Maxim Nikulin writes:
>
>> On 14/04/2021 02:08, Juan Manuel Macías wrote:
>>> The situation is that with locales configured for Spanish from Spain
>>> (en_ES.UTF-8) the list is not ordered correctly, unless those three
>>> spaces from org-sort-remove-invisible are removed. But I couldn't say
>>> why or if that would be appropriate as a patch...
>>
>> Did not you get a warning like the following one?
>>
>> (process:220): Gtk-WARNING **: 15:17:45.066: Locale not supported by C
>> library.
>>       Using the fallback 'C' locale.
>>
>>> Regarding the collation rule of forms with emphasis, at least in Spanish
>>> these should come after the non-emphasized forms. I don't know if this
>>> is a general rule also for other languages (at least it seems more
>>> natural that the forms without emphasis are placed before).
>>
>> LANG=es_ES.UTF-8 \
>>   emacs -Q -L ~/src/org-mode/lisp/ -L ~/src/org-mode/contrib/lisp/ \
>>   list.org
>>
>> - a
>> - /a/
>> - v
>> - /v/
>>
>> So it works accordingly to your expectation. Emacs 25.2.2,
>> Ubuntu-18.04 container.
>>
>> I have generated es_ES.UTF-8 locale using
>>
>>     dpkg-reconfigure locales
>>
>> Depending on linux distribution you run, the locale may be ready to
>> use or not. I tend to think that in minimal environment of virtual
>> machine it was missed.
>>
>> I had an idea to add a test for sorting of items including emphasized
>> ones but test-org-list/sort forces C locale. Maybe it was done to
>> avoid failures due to missed locale.
>>
>>
>