Re: Unhappy about tools and update policy

On Sat, May 28, 2011 at 10:03 AM, Kathiravelu Pradeeban
<kk.pradeeban@gmail.com> wrote:
> Hi Chris,
> Thanks a lot for willing to help AbiWord on this. Apologies everyone
> for the delay in the response from our end on this regard.
>
> On Sat, May 28, 2011 at 7:13 PM, Martin Sevior <msevior@gmail.com> wrote:
>>
>> Hi everyone,
>>
>> I think we still have an 2.8.7 release due. uwog just needs to decide to do it.
>>
>> After that it's full throttle on 3.0.
>>
>> With regards to the translation tools. I think we could gain from the
>> services offered by Chris and the OLPC team. I get the feeling that
>> uwog is kind of overloaded right now. If another committer could step
>
> True. Marc (uwog) is overloaded now. I would be happy to work with
> OLPC team on this from AbiWord' end. Nevertheless, I would also like
> hear from Marc on this.

Pradeeban,

I'd be happy to work with you on this collaboration. Marc (uwog) is
well known around the Sugar community (often hanging out on our IRC
channels) and I respect the fact that the is busy. I believe we will
be able to resolve most issues without bothering him, although I too
would value his feedback.

It's a holiday weekend here in the US (Memorial Day), I may not be
able to work on this too much until Tuesday, but here are some initial
thoughts.

The first issue to resolve is the mapping of languages and ISO-codes.
Our Pootle instance may need a little adjustment to add some new
languages/codes. We typically use language (ISO-639-2 or ISO-639-3)
and only split for locale when needed (for instance our large South
American user base is in favor of keeping a single lang-es project)
whereas AbiWord codes typically include the Country/Region code like a
glibc locale, even if there is not a second region for the primary
language.

I've done some analysis (see attached spreadsheet), and it appears the
vast majority of these are easily resolved (137 codes total), with
only 28 codes requiring further review and discussion with localizers
and developers to resolve in a mutually satisfactory manner.

Even among these 28, some will be simple (e.g. different PO files for
Cyrillic or Latin orthography), I would just want their input first as
these languages are less familiar to me. A few may require some
consensus building on whether a split is truly required and meaningful
(e.g. a few of the many English variants), justifying the additional
burden on L10n maintainers, although I am inclined to err on the side
of respecting regional autonomy. Some like Quechua and Aymara are
truly challenging, but I've already been talking to your
Quechua/Aymara team (through Amos Batto) as Sugar Labs / OLPC faces
the same challenges with properly representjng and localizing these
indigenous languages in a region with significant XO deployments and I
am sure that a suitable resolution can be achieved with localizer's
input.

57 entries = Class 0, Essentially one-to-one correspondence, simple
upload of existing PO from AbiWord

7 entries = Class 1, Single code on Pootle, multiple in AbiWord, resolve

21 entries = Class 2, Code does not exist in current form on Pootle,
add new or resolve multiple code mapping

80 entries = Class 3, Code exists in Pootle, not yet in AbiWord,
nothing to migrate, populate with template

I look forward to working with the AbiWord team to our mutual benefit
and the benefit of our customers and communities.

Warmest Regards,

cjl
volunteer Sugar Labs / OLPC / eToys Pootel admin