Re: [PATCH] localedata: Reorganize Unicode LC_CTYPE inclusion.

[Carlos O'Donell <carlos at redhat dot com>]

On 10/26/2017 02:35 PM, Rafal Luzynski wrote:
> 25.10.2017 18:21 Carlos O'Donell <carlos@redhat.com> wrote:
>>
>>
>> On 10/25/2017 03:03 AM, Rafal Luzynski wrote:
>>> 24.10.2017 00:33 Carlos O'Donell <carlos@redhat.com> wrote:
>>>> On 10/23/2017 03:21 AM, Rafal Luzynski wrote:
>>>>> I will appreciate if you fix these problem on your own, the last problem
>>>>> at least.
>>>>
>>>> I will fix this immediately and commit the changes. Thank you again for the
>>>> review.
>>>
>>> I can't see your commit. Please do these changes or ask other people
>>> (including me, of course).
>>
>> Committed. Just needed to test a few more things. My apologies for the delay.
> 
> Thank you.
> 
>> I'm happy to review any patches that fix mistakes I've made, please don't
>> feel that you need to wait for me either. As a subsystem maintainer you can
>> post a patch with "[committed]" status that fixes my mistakes, and you may
>> do so immediately :-)
> 
> OK, I'll be braver next time.

No worries.

> However, I see that the patch you have committed [1] is different from
> the patch you have proposed. [2] All those "category" lines are still
> there, also "category ...;LC_TYPE" line appears twice. Can we remove them?
> What do you think about the attached patch? Sorry, I'm not sure so I need
> a review this time.

As subsystem maintainer for localedata you do not need review. You can assume
consensus and fix the file directly.

You are correct that the category lines are in error, they should be removed,
I didn't notice them in my final regeneration.

However! The category line generated by localedata/unicode-gen/gen_unicode_ctype.py
is also wrong.

It is a static 'unicode:2014', and this is invalid.

ISO 30112 standard defines the valid values for the category.

There are only 4 valid values:
"i18n:2012"
"i18n:2004"
"posix:1993"
and they state the conformance of the category itself.

Next step:

* Remove the category entries in i18n_ctype except for LC_CTYPE.
* Adjust the category entry for LC_CTYPE to use "i18n:2012"
* Adjust gen_unicode_ctype.py to output "i18n:2012" instead of the invalid
  "unicode:2014"

We should make localedef validate the conformance of this value, but that's
another patch.

I will assume you will make the changes unless you hand them off to me.

-- 
Cheers,
Carlos.