Re: [PATCH 3/5] localedata: CLDRv28: update LC_ADDRESS.country_name translations

[Mike Frysinger <vapier at gentoo dot org>]

On 09 Feb 2016 16:01, Carlos O'Donell wrote:
> On 02/09/2016 03:00 AM, Mike Frysinger wrote:
> > On 09 Feb 2016 09:25, Marko Myllynen wrote:
> >> This series is a great initiative, thanks for doing this! I wonder how
> >> have you generated the patches, would it be possible to script any of
> >> this so that in case CLDR changes in the future we could automatically
> >> update glibc locales accordingly then?
> > 
> > yes, i've written a python file that downloads the CLDR db, loads the
> > locale files, and then merges them and checks for differences.  i'm in
> > the process of trying to extract as much as possible from the CLDR db.
> > you can find my in-progress work attached, but i'm not posting it for
> > review yet ;).
> > 
> > i just run it like:
> > $ ./cldr.py locales/*
> 
> So basically I think that reviewing your patches is only mildy useful
> busy work.

i split up the changes to make them easier to review w/out getting
fatigued and to make it easier to pick out errors.

> What's *really* cool is getting your scripts to a point where
> they can be used like the Unicode scripts are used, a single directory
> and a Makefile that can be run to update the files in question.

it's getting closer.  some of the fields are harder to look up than
others.

> I think we're going to have to checkin the CLDR db data to ensure we
> can reproduce the results?

the script lists the original upstream URI and will download it on the
fly.  since the db is 13MB compressed and 90MB uncompressed, i'm not
sure it's something we want to add to the repo.  even if we try to drop
unneeded files, we're still talking 40MB+.

if we're concerned that CLDR might go away, we could always mirror the
svn repo into sourceware.org git.  i don't think it's a problem myself
as we're already checking in the generated localedata/ files.
-mike

Attachment: signature.asc
Description: Digital signature