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

[Mike Frysinger <vapier at gentoo dot org>]

On 11 Feb 2016 10:24, Florian Weimer wrote:
> On 02/11/2016 05:27 AM, Carlos O'Donell wrote:
> > On 02/10/2016 03:12 PM, Mike Frysinger wrote:
> >> then it would be obvious what version of CLDR was used to update the
> >> locale.  the downside is that the file isn't 100% sourced from CLDR,
> >> so it seems like clobbering all the fields is wrong ?
> > 
> > Per FSF statement [1] the locale files are not copyrightable so IMO
> > attribution matters only so much as we care to thank the previous
> > authors for their work. Such previous authors already have attribution
> > in the Changelog, and IMO need not have any more attribution in the
> > source file, just like we don't use "Contributed by" anymore.
> 
> unicode.org claims copyright on CLDR data:
> 
>   <http://unicode.org/repos/cldr/trunk/unicode-license.txt>

perhaps on all the files and their collection and additional data, but
you can't copyright facts.  i'm not committing their files.

> The terms do not appear to be too onerous, but I would recommend to
> obtain FSF (and internal) sign-off before incorporating data directly
> from CLDR into glibc.

covering our bases makes sense, but i don't see a problem here.  i'm
querying their db for an uncopyrightable fact ("what is the weekend
start day -> saturday") and then merging the answer "saturday" into
the data files on our side.

also, we're already doing this today: localedata/unicode-gen/ parses
the unicode.org databases to generate the list of valid characters.
-mike

Attachment: signature.asc
Description: Digital signature