Re: Complaint of the Slovak coordinator

[Marcel Telka <marcel telka sk>]

Hi Johannes,

Thank you for your view.

On Thu, May 13, 2010 at 05:33:09PM +0200, Johannes Schmid wrote:
> Hi!
> 
> (This is my personal view, if solving this should prove difficult we
> might need a meeting of the coordination team to discuss this in
> detail.)
> 
> After reading through (the English part of) the mails I came to some
> conclusions what is wrong in this team:
> 
> * Trust: To me it seems that there is a lack of trust between the team
> members and the team coordinator. I think when you have some reviewers
> you should trust in them to make a final review. People will always make

You must distinct between mistakes everybody is doing and between lack
of knowledge/skills. :-(

> mistakes but it's better to "release early, release often". There are
> scripts that usually fix all the release critical bugs. If you find bugs
> later you can still fix it. But as long as a translation isn't public

That's true. But if you have users, it does not automatically mean
you'll have bug reports. Yes, you'll get some of them but it is not
clear if it would be enough. From the past I know most of translation
bugs we found were found during the translation update by translators or
during the review. Our user base is far smaller than German or even
Czech.

> there won't be any users and thus no bug reports. So, I would encourage
> you to drop the final review stage (now).

To have this done I would have to set up more strict rules to become a
reviewer. I opted for different approach: to allow all members to try to
do the reviewer's job to see how it will go.

This allows us to catch at least some bugs by current reviewers. And
this brings up a chance for potential good reviewer to show and improve
his work. In case I would allow to review only members to whom I can
trust the proces would slow down even more.

> 
> * Positiv attitude towards newcomers: First, nonetheless how it is
> implemented, a formal mail is not the way to go. We already place a good
> amount of hurdles in the way of new translators by having to get used to
> git/damned-lies, po-files, etc. This is not like launchpad where
> everyone can start translating and that's done on purpose.

This is not like launchpad. I agree. But the difference is not very big.
To have started a translation you just need to register to the Damned
Lies, join the team, reserve for translation and submit a po file. You
can do it in few minutes. I do not know how launchpad works, but I think
it would be similar easy.

> That means that everybody who got that first step will already have at
> least some understanding of the workflow and put effort in solving a
> steep learning curve.

Unfortunatelly, this is not true. Sometimes the workflow knowledge of
new members is poor.


We had several people who joined the team via Damned Lies and nothing
else. After some time I asked them whether they really wants to became a
members and help with translations and asked for subsequent
actions tobe done (bugzilla account, mailing list subscription).
No answer for more than a week or two. So I removed such people
from the team and sent a mail stating the reasons. After that
they just replied they are ok with the removal because they have
no free time to do anything with the translations now. This is
quite often. It is easy to have dozens of team members listed at
the team page...

> Of course translation quality might be poor in the beginning but

Yes. This is the reason why we are doing reviews.

> everybody was a newcomer first. Usually it's good to setup some kind of
> rules for consistent translations were people can stick to. 

I agree. Now, we are in process to have such rules set up.

> 
> * Long-term vision: You will almost never find someone who will clearly

Sorry, this was misinterpreted. The rule is that a potential translator
for a module should say: Ok, I'll maintain this module. Once his
interest for the module is lost he can easily say: Ok, I'll no longer
maintain this translation, it is free for someone else. There is no
requirement to became a translator for several years.

The assigned translator have several pros:
- the translation will be easily maintained self consistent
- when a bug is filled it is easy to realize who is the proper person to
  whom the bug should be assigned

> state that he will be able to maintain a translation for the next few
> years. This is a volunteer project and life changes sometimes. But it is
> not a problem when different people translate the same module, neither
> for consistence nor for quality if done right.

It might be not a problem, if done right. True. We are finding the
proper approach how to do that. As for now the "assigned translator"
approach seems to be ok. Peter's proposal was not to remove the
"assigned translator" practice. He proposed just some changes to the
assigning process. So I think most of the team is ok with the "assigned
translator" approach in general.

> Reviewers will notice when there is an inconsistency within the strings
> and are able to point that out. I think most teams have no problem in
> sharing modules between different translators.

True. This is why I am doing now final review for all modules. I hope
this will change soon.

> => Assigning modules to a single person without giving others the chance
> to step up is a bad thing.

I disagree. It depends. Other translators can select other modules, so
the rules are same for everybody.

Some people does like to feel that this module is their work and they
are responsible and proud for them.

> 
> * Personal differences: This happens, accept it. We don't need to be
> best friends all and we sure aren't. This is about getting work done
> which definitely fails here as the numbers show. 

I agree. Unfortunately we spent a lot of time in discussions where I
lost time to do some useful work :-(.

> 
> So there is a problem and it is not a good idea to say everything is
> fine. As well as it is not true that everything is wrong.

I agree. I never said everything is ok. But I cannot accept that
everything is wrong, as some people are trying to present.

> 
> Marcel, I would love if you could point out how to solve this problem
> with real actions. You only talked about the past till now, please talk
> about the future.

In the team we work on improvements. We did something in near past
(registration form, more reviewers involved), we are working on
something now (translation rules, wiki pages improvements). For future
we will to create more and more translation rules, I'll work on my
translations and final reviews. And finally, I hope we will have shortly
at least one well skilled reviewer who will became another "final
reviewer".

I'll try to push that translations are reviewed by more than one
reviewer. This will catch more bugs, so my "final review" would be
easier, simpler, and faster. And one day (as I said above, I hope
shortly) my final review will not be needed at all.

> 
> Thanks and regards,
> Johannes
> 
> BTW, Marcel is missing as committer here: http://l10n.gnome.org/teams/sk

I am corrdinator, so I am commiter, reviewer and translator too. This is
how Vertimus shows the roles.


Thank you.

-- 
+-------------------------------------------+
| Marcel Telka   e-mail:   marcel telka sk  |
|                homepage: http://telka.sk/ |
|                jabber:   marcel jabber sk |
+-------------------------------------------+