Pootle instance outside of Mozilla

[Eduardo Trápani]

There is something I don't understand. Why can't somebody set a Pootle
instance outside of Mozilla?

For ages I've been doing my localization work with software I had
written myself (trying at a Pootle/Pontoon kind of thing after a project
with Dewi) and also with a text editor or virtaal for iOS, depending on
the needs. And then I simply used svn, mercurial or git to push the
changes from my local copies. Only lately I've been using Pontoon
because it made it a lot easier (Pootle would have been the same) for
contributors.

For me, *that* is a good chunk of what I understand by openness. We can
use whatever we want to localize, over open systems/protocols
(git,hg,...) on known open formats (xml, xliff, properties,...) as long
as we follow the rules. That's great. That gives room for
creation/adaption of new solutions: Pootle, Pontoon and other in-house
systems like the one I use, that can be helpful but might never see the
public light.

So, if svn/mercurial/git accounts/access are still assigned to
localizers, then a Pootle instance outside of Mozilla could serve those
communities that don't want to migrate away from it. If for example .ftl
doesn't work, somebody will try to code it and contribute the software,
that's how open source works. If nobody codes it, for a while some teams
might have to resort to other tools or workflows for certain files,
projects, or make hacks. Or turn to Pontoon.

For those that love Pootle and, most importantly, are productive with
it, I would suggest to start thinking on how to migrate the current
instance away from Mozilla and start looking a new home, or many homes
for it. Find out what it's needed to run it, to synchronize, etc. A
community of "drivers" could push the changes to the repositories if it
need be, to keep users away from that part of workflow. It could work. I
offer some help for that.

No need to leave or lose all the time, training and know-how, all the
love each one of us puts into her/his locale. You can find a way to keep
using Pootle, not by asking Mozilla to keep it and pay for it (that is
pretty much their call), but making it work equally great outside
Mozilla using the open tools and formats available to us. If Pootle can
keep up, then it will survive (for Mozilla l10n) on its own merits and,
who knows, sometime in the future it might even find it's way back to
the Mozilla core.

Now, if Mozilla decided not to allow direct access to the repositories
anymore, efectively forcing Pontoon on everybody as the only possible
localization tool ... that would be another issue. But I don't think
that's the case.

Jeff, are repositories still going to work as they have always been?
Could we install an external instance and sync over svn/hg/git either
directly from Pootle or via a localizer?

Eduardo.

[Jeff Beatty]

Hey Eduardo,

Thank you for the thought and detail of your email. I hope that I can
provide an equally thoughtful answer.

On Wed, Jul 26, 2017 at 2:10 PM, Eduardo Trápani <etra...@gmail.com> wrote:

> There is something I don't understand. Why can't somebody set a Pootle
> instance outside of Mozilla?
>
> For ages I've been doing my localization work with software I had
> written myself (trying at a Pootle/Pontoon kind of thing after a project
> with Dewi) and also with a text editor or virtaal for iOS, depending on
> the needs. And then I simply used svn, mercurial or git to push the
> changes from my local copies. Only lately I've been using Pontoon
> because it made it a lot easier (Pootle would have been the same) for
> contributors.
>
> For me, *that* is a good chunk of what I understand by openness. We can
> use whatever we want to localize, over open systems/protocols
> (git,hg,...) on known open formats (xml, xliff, properties,...) as long
> as we follow the rules. That's great. That gives room for
> creation/adaption of new solutions: Pootle, Pontoon and other in-house
> systems like the one I use, that can be helpful but might never see the
> public light.
>

I certainly see your point and agree, however, if the process and
technology of the FOSS software your localizing changes rapidly, tools that
don't have consistent investment or support won't be able to keep up. At
that point it's not about openness but about technical requirements to
participate effectively. I'll address the technical requirements below.

Right, that is not the case. Those working offline and committing to VCS
are able to continue to do so and Pontoon will sync with both the en-US
repo and the locale's repo. This ensures that whether you contribute
through Pontoon or by committing directly to VCS, your contributions are
synchronous between the two and attributed appropriately.

>
> Jeff, are repositories still going to work as they have always been?
> Could we install an external instance and sync over svn/hg/git either
> directly from Pootle or via a localizer?
>

Coming back to those technical requirements. Because of all of the changes
to the localization process over the years (and those to come), we have to
assume that all tools used for localization at Mozilla are capable of:
1) Bidirectional sync with version control repositories (e.g., allowing
us to migrate content from the old format to FTL seamlessly for localizers,
or allowing localizers to push to version control from another tool without
causing merge conflicts or backing out what one localizer committed when
another commits).
2) Committing contribution attribution appropriately to version control
repositories.
3) Support for l20n (.ftl format), including a GUI to facilitate the
use of ftl syntax in translated strings.
4) Support for the upcoming central project configuration (part of
single-repo, cross-channel plan).

This list will be amended over time to support any changes to the
development processes of Mozilla projects. The important thing here is that
change is coming pretty fast. Now that localization is taking place on
Nightly, new source strings are landing every 2-3 days. We expect
cross-channel to be complete for Firefox 57 and we've gotten the go ahead
to land the core elements of l20n in Firefox, making it smooth sailing for
implementing it fully as early as Firefox 58. Should such an instance of
Pootle exist, it would need to keep up with these changes in order to
facilitate localization, and adapt to future changes that come down the
pipeline. For these reasons, we strongly suggest that localizers to move to
Pontoon as the path of least disruption to their localization contributions.

I hope this isn't a ramble but an informative and valuable response. Happy
to answer any other questions (or Pike might answer anything more
technical).

Thanks,
Jeff

>
> Eduardo.
>
> _______________________________________________
> dev-l10n mailing list
> dev-...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-l10n
>



--
Thanks,
Jeff

[Rhoslyn Prys]

Thanks for this more detailed reasoning for moving to Pontoon. I can
understand your need to control the localization process in line with
future requirements. This is what other software companies with L10N
requirements do, I get feedback on their processes (I've not signed any
NDAs!) and the messages I get is that Pootle is far better than the
other systems used. The main difference for Mozilla is that other
companies pay translators and Mozilla on the main depend on volunteer
translators, so there should be plenty of spare cash to provide an
instance of Pootle for those who prefer to use it.

My concerns regarding Pontoon is that, although it's had plenty of time
to catch up it's still not as good as Pootle and I've little confidence
that it will ever get there and certainly not before the end of August.
If it's a matter of using L20N I'm far from convinced of it's usefulness
and would be happy to forgo it. It has the potential of being a right pain.

Please review your decision.

Rhos

[Eduardo Trápani]

Thanks Jeff! Everything is a lot clearer now.

> dev-...@lists.mozilla.org <mailto:dev-...@lists.mozilla.org>
> https://lists.mozilla.org/listinfo/dev-l10n