[Systems] [support.osuosl.org #5846] Sugar Labs: Hosting for Pootle

[Chris Leonard]

On Sat, Jan 31, 2009 at 3:00 PM, Luke Faraone <luke at faraone.cc> wrote:

>  On Sat, Jan 24, 2009 at 2:36 AM, Bernie Innocenti <bernie at codewiz.org>wrote:
>
>> This server has just transitioned into production a few days ago, but
>> we could host stuff like Pootle there too.
>
>
> Alternatively, something we can do that requires no hosting effort or
> resource consumption: Use Launchpad.net's Rosetta.
>
> Rosetta is hosted by Canonical with a team of full time sysadmins. They are
> planning on opensourcing it<https://dev.launchpad.net/OpenSourcing#schedule>along with the rest of launchpad on July 21, and already support the import
> and export of POT files.
>
> We'd be able to make use of this without any other commitments to Canonical
> infrastructure, and Canonical has pledged to continue to provide free
> hosting to open source projects, and to not roll out premium features only
> available to paying customers.
>
> Or we can host and manage Pootle ourselves.
>
> Luke,

There are good reasons why many of the Ubuntu upstreams maintain a separate
localization infrastructure closely coupled to their repo infrastructure.
While the latest Launchpad 2.2.1 release (only days old) does contain
some improvements in the Rosetta localization tool that seem to make it more
workable for upstreams, I am not convinced that abandoning a
locally-controlled Pootle instance would be a desirable option at this point
in time.

see:  https://launchpad.net/launchpad-project/+announcement/1904

In particular, I believe that abandoning a local Pootle instance and relying
entirely on Launchpad Rosetta (in the absence of a migration to Launchpad as
the primary repo) , would be seriously disruptive of the work of the Sugar /
OLPC localization community as it currently exists.  In addition, it would
mean surrendering a certain level of control, admittedly in exchange for
a lowered administrative burden, but I'm not sure that that payoff would
be adequate to make up for the many  potential negatives of  such a drastic
change (less control, learning curve on new tools for both admins and users,
etc.).  This calculation may change in time, but at preseent, I do not see
Launchpad as an acceptable alternative to a locally maintained Pootle
server.

One could imagine a scenario in which one included a Rosetta-based
translation capability with in the Launchpad sugar packages *in addition* to
a local Pootle instance, but that would also introduce additional work on
keeping them synchronized that would make it a less-than attractive option
until there is some long-term stabilization of the local Pootle situation,
which is why there have been discussions exploring the establishment of a
Sugar Labs hosted/controlled Pootle instance and an orderly migration from
the current OLPC hosted instance, to be coordinated with an upgrade of the
Pootle version.

Current plans lightly sketched out in this message to L10n list.
http://lists.laptop.org/pipermail/localization/2009-January/001915.html
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.sugarlabs.org/private/systems/attachments/20090131/ece466b9/attachment.htm