On Fri, Sep 16, 2011 at 10:10 AM, Hannes Magnusson
<[email protected]> wrote:
> On Thu, Sep 15, 2011 at 23:44, Pierre Joye <[email protected]> wrote:
>> On Thu, Sep 15, 2011 at 11:42 PM, Stas Malyshev <[email protected]> wrote:
>>
>>> The problem is data may vary by CLDR version, etc. and keeping pace with all
>>> variants on all platforms, etc. will be a nightmare.
>>
>> To have a couple of versions, say x.y, x.y+1 sounds reasonable and possible.
>
> Its not our responsibility to test the 3rd party library data, thats
> for intl to test.
It is not about validating the correctness of ICU itself but our
binding or APIs usage. And to do that we have to deal with the ICU
results.
--
Pierre
@pierrejoye | http://blog.thepimp.net | http://www.libgd.org