Re: [PATCH v3] Add new C.UTF-8 locale (Bug 17318)

[Carlos O'Donell via Libc-alpha <libc-alpha@sourceware.org>]

On 4/28/21 1:36 PM, Joseph Myers wrote:
> On Wed, 28 Apr 2021, Carlos O'Donell via Libc-alpha wrote:
> 
>>>> diff --git a/localedata/locales/i18n_ctype b/localedata/locales/i18n_ctype
>>>> index c63e0790fc..c92bb95148 100644
>>>> --- a/localedata/locales/i18n_ctype
>>>> +++ b/localedata/locales/i18n_ctype
>>>> @@ -26,7 +26,7 @@ fax       ""
>>>>  language  ""
>>>>  territory "Earth"
>>>>  revision  "13.0.0"
>>>> -date      "2020-06-25"
>>>> +date      "2021-02-17"
>>>>  category  "i18n:2012";LC_CTYPE
>>>>  END LC_IDENTIFICATION
>>>
>>> Those date changes seem spurious.  Is this no-op file regeneration
>>> really needed?
>>
>> The protocol is:
>>
>> cd localedata/unicode-gen
>> make install
>>
>> The spurious regeneration is not needed, but it's easier to run the
>> above commands. It gives a date for the last generation for all files
>> consistently.
> 
> Is it required to have a date in the file at all?
> 
> For anything that's generated during the glibc build and ends up 
> installed, we'd avoid having such dates to allow for reproducible builds.  
> As this is a maintainer generation of something checked in as a source 
> file, rather than regenerated from makefile dependencies in a normal glibc 
> build, that doesn't strictly apply, but I still think it would be a good 
> idea to avoid having dates or other such non-reproducible output in 
> maintainer-generated files as far as possible.
 
There turns out to be a case where the dates are important.

The LC_IDENTIFICATION for the locale, particularly the 'date' field.

Now the entirety of LC_IDENTIFICATION is a GNU extension that is metadata
about the locale itself and has no strongly documented meanings AFAIK.

My plan is to use the Unicode release date as the consistent date.

Every time we update Unicode to a new version we would see a date change,
but subsequent regeneration would not see any changes.

-- 
Cheers,
Carlos.