On 2018/10/05 20:48, Jean-Baptiste Holcroft wrote:
On 05/10/2018 10:40, Noriko Mizumoto wrote:
> Hi Jean Baptiste
>
> Thank you for working hard for the team. Now I am feeling fear that
> many important things are moving forward in the dark without any
> interaction with many of translators here. Could you mind pause a bit?
Yes, of course!
I'm afraid this is like any project, if the project is working on the
internationalization process and we are interested in, we'll find
information in this project mailing lists/tools. Sometimes we have
dedicated people for cross-project coordination, but time for
coordination really is an issue... I personally tried but don't have
much time for now.
It is not something like add-on feature or package starting from
small-mid size to current platform but possible replacement as larger
scale to happen in short window, which directly impact entire
translation team activity, as well most of the parts of fedora project
which use translation from us. In this way, I see it, not like any other
project.
> First, Antra can't deal with zanata? If it can, it is best to keep
> using zanata for a while until everything to be cleared and confirmed.
> if it can't, any background what made docs team to move from publican
> to antra?
Antara is a documentation tool, it isn't connected to any translation
platform.
I don't feel like to be the right person to explain why the docs team
moved to Antara.
They decided it publicly, published it, it works. Is there an issue with
that?
Yes, it is critical issue. As you said, Antra isn't connected to any
translation platform, then we are no longer able to translate release
notes and any other documents via Zanata, our platform. What is going on?
> Second, I would like to know clear reason/cause if any move our
> platform to somewhere is really required?
>> If you go in the l10n repositories, you'll see Weblate commits. I
>> used it for my tests and will propose the community to take the
>> opportunity to test the tool for a few days. Anyway, we made sure the
>> process is platform agnostic.
>>
>> If you want to have a close follow up of the progress of this, go to
>> the fedora docs on telegram or irc. I'll come back here to ask for
>> testers.
>
> Before testing, as said, we should go through the steps of "clarify
> the situation" and "identify whether we move or stay zanata". In case
> we DECIDE(1) or NEED(2) to move, then we can move forward to PLAN it,
> such as "list all possible tools (incl. weblate)", "evaluate/test the
> tools listed", "vote", "approach devel team", etc.
We can imagine multiple ways to discuss the point, but let's not take it
as a long-term decision, but more as a short-term opportunity to see
something else.
Having time and skills to tests such things isn't so easy. Only
gathering translation team leader in a meeting is really hard. We often
had more skilled internationalization people in our meetings than
translators.
Did we? my memory says that majority of our irc meeting attendees were
translators, having a few i18n and other team members who were willing
to support us.
The hosting of Zanata in our infrastructure is in discussion since
the
beginning, and never worked. So having skilled administrators also is
really hard.
I never know there was discussion of hosting zanata in our infra
(supposed fedora-infra). It was part of the plan as the condition that
zanata provides their own infra for fedora-l10n because fedora-infra
can't afford.
I started to get interest in learning how to host Weblate more than a
year ago.
If anyone else with a ready to use translation platform is really
welcome to demonstrate it with whatever subject he/she wish.
This is testing, so
I have no intention to against your interest to weblate, rather I see it
can be good option. But why you so rush into testing?
If I go for shopping my car, I would compare as many cars as possible
before having test drive with specific car.
I am asking here, we are even not knowing whether we should go for
shopping car while we have car called "zanata".
> Please let me press this point, the hardest part in entire process, is
> convincing devel team. There were 900 translators of 90+ language
> teams, 70 packages with associated developers, 29 books with
> associated authors and web team, all had to agree and take action for
> the move last time. Especially every developers have to setup their
> repository from scratch again. We lost certain number of translators
> and developers when moving from transifex to zanata.
> Therefore, we better gather all the necessary information first
> (rationalizing phase) and develop a plan (planning phase) in written
> form, then implement the plan (action phase) in order to minimize any
> loss.
Well, we should really plan a migration if we had to change. But here I
propose to test another tool to see if we like it or not.
What to you feel, are the reasons of the loss related to the planning?
Unfortunately not feeling, but fact.
I experienced this three times, first was elvis to transifex-internal,
second transifex-internal to their external, and then third transifex to
zanata. Just to compare the number of registered translators from my
memory, 4000 translators registered with trans lists was decreased to
2000 at the end of second, then third less than a thousand.
The move will anyway lose/put behind some people even it is agreed
decision, this is like painful overhead (sorry not sure this term is
appropriate). We still have kept hearing the claims from some packagers
"I didn't know at all about zanata", even it was 3 years ago. The loss
can happen everywhere, and we pay for it.
Shall we start evaluating Zanata sustainability or assessing the
severity first, in order to shape our future outlook?
noriko