Re: short day translations vs cldr entries

[Marko Myllynen <myllynen at redhat dot com>]

Hi,

On 2016-02-09 16:22, keld@keldix.com wrote:
> On Tue, Feb 09, 2016 at 11:10:37AM +0200, Marko Myllynen wrote:
>> On 2016-02-09 09:53, Mike Frysinger wrote:
>>> should glibc localedata conform entirely to what is in cldr ?
>>
>> I think most of the localization experts and national agencies are
>> contributing to CLDR so in general I think the answer would be yes.
>> However, we probably should not blindly copy what is in CLDR but check
>> the differences before merging and if we think glibc is more correct
>> than CLDR then submit change requests to CLDR. In the end it would be
>> very beneficial to have glibc and CLDR in sync. Of course there might be
>> some corner cases where this is not possible, we might need to prepare
>> for some expections (and some data is only available in glibc).
>>
>> I'm not sure about the exact cases you asked wrt German/French short
>> dates but I think in general your script seems to be doing the right
>> thing so far given that there has been no changes seen yet for fi_FI
>> which AFAIK should already be in sync between glibc and CLDR.
> 
> My understanding of contributers to CLDR is that they are only a handful of experts,
> and that only a few national agencies are contributing, including Finland and India.
> 
> I believe we have more contributers in glibc-i18n than there are contributers to
> CLDR.

I'm not sure about that - looking at the release notes and the changelog
for their latest release give a sense of steady progress:

http://cldr.unicode.org/index/downloads/cldr-28
http://unicode.org/cldr/trac/query?status=closed&status=reviewing&resolution=fixed&milestone=28&group=component&max=999&col=id&col=summary

> Furthermore CLDR has been unresponsive to some requests, including Danish
> requests. So I would not recommend that we just follow what they say. 
> We should take input from them, but not automativcally align.

Do you know why those requests were not resolved, do you have references
to those CLDR tickets? And yes, automatically and blindly syncing might
be a step too far at least at this stage but pulling in fixes should
happen much more frequently. Perhaps it might be possible to document
known deviations to certain extent or even extend Mike's script to omit
those cases (similar to the exceptions in the scripts used now to pull
in the Unicode data). But I think we need to gain more understanding of
the current situation before committing to anything.

Thanks,

-- 
Marko Myllynen