Re: Complaint of the Slovak coordinator

[Jiri Eischmann <jiri eischmann centrum cz>]

Marcel Telka píše v Čt 13. 05. 2010 v 12:44 +0200:
Hi,
> 
> Some things I explained in my previous post, so I'll not duplicate my
> response here. If I overlooked something (and thus deleted it), please
> let me know, I'll add my comments.
> 
> On Thu, May 13, 2010 at 01:08:52AM +0200, Jiri Eischmann wrote:
> > JOINING THE TEAM: In the Czech team, you just need to sign up on
> > Vertimus and make a modul reservation. It's also recommended to send an
> 
> We require following for new members:
> - join gnome-sk-list
> - register to bugzilla
> - register to Vertimus and join the team
> - sent an email (with a template available)
> 
> Before a translator can start to translate he should sent an email with
> the request. In less than one working day I either assign the module to
> the translator (most cases) or explain why it is not possible, or why it
> is not good idea.
> 
> > e-mail to the mailing list to check if someone's accidentally
> > translating it. In the Slovak team, you have to send a formal e-mail in
> > which you have to write your real name (tell open source people about
> > privacy...). You have to ask for moduls and many of them are not
> > available even though no one has touched them for months.
> > CHECKING TRANSLATIONS: In the Czech team, a translator uploads a
> > translation and a corrector checks it. If he finds any incorrect
> > translations he corrects them and if he thinks there are serious he
> 
> If you have skilled translators and reviewers you can do that. This is
> my dream for our team too.
> The problem is that we don't have skilled translators as well. When will you be ready to accept the reality? This is not proprietary software where people translate software as a job. People we deal with are volunteers and do it in their spare time. The circulation of people in open source projects is much faster. In the Czech republic, we give newcomers much higher opportunity. They are learning by doing (by translating in this case). There is a guy in the Czech team who's been a member less than a year. For that time, he has translated so much that I was shocked (in the positive way). He doesn't have to wait months until his translations get to git. Otherwise he would leave the team very soon. I'm pretty sure in your team he would be considered unskilled and needed to be trained. The reality is that our translators don't have time to learn for months or years. If you don't accept it it will never get better.
You might argue that those more free rules lead to lesser quality of translations. I'm not saying that the Czech translations of GNOME are perfect, but the users' feedback is positive and localization of GNOME-based products (e.g. Ubuntu) is reviewed very positively even by a non-Linux media. And Slovak translations? Users are using English or Czech ones instead. Can you get a worse answer from users?

> provides feedback to the translator. But he doesn't return the
> > translation! The time between uploading and submitting is about a few
> > days, 2 weeks max. In the Slovak team, if a corrector finds a problem he
> 
> This is not general rule. If reviewer finds easily fixable problem (like
> typo or obviously invalid translation where there is clear way how to
> transllate it) he will fix it by himself. If the problem is not easily
> fixable, or the proper translation is unclear, the message is marked
> properly (with comments) and left for translator to do his job.
> 
Compare logs under moduls in the Czech team and the Slovak team. Some translations are returned again and again. Why don't you make unclear or incorrect strings fuzzy and commit the translation to git? A translator can fix it later and users can use at least some translations. 

> This way works all reviewers in our team, not only me. To be exact I do
> not review newly uploaded translations for some time. This is job for
> our 4 reviewers.
> 
> > returns a translation to a translator. This happens again and again
> > until it's perfectly correct or more likely the translator gets
> 
> Again, "perfectly correct" is not true.
> 
Well, if you want to base your arguments on word nuances... we can call it a quality level
which is discouraging for volunteers.
 

> frustrated and gives up. No surprise that there are many translations
> > waiting to be submitted for months and I think some of them will never
> > be submitted.
> 
> If a translation is never submitted then it cannot reach the git. I do
> not understand this, sorry :-(.
> That's a very smart observation ;-) Yeah, if translations get stuck in Vertimus, they can't reach the git... until some one who's got the right submits them.

> OTOH, we have examples where translators complained about too loong time
> between first submission and the final commit. When I sat down with the
> translator and started to anlyze the Vertimus data we found that most of
> the time was spent by either translating the module or reviewing it.
> Overhead generated by my final check was about 15 %. After that the
> translator admitted, that his copmlains were about feeling, not real
> data.
> How can translating slow down the process between uploading a translation into Vertimus and getting to the git? If you don't accept translations until they are >90 % it's the 60% of Empathy case all over again. Instead of providing users at least a portion of a translation you don't provide them anything. Not only for a few months, but so far for years. For example I translated the Evolution documentation for a year and half (it has more than 2000 strings and strings in documentations are longer than in apps). If I had had to wait almost two years until users can use my work I would have probably given up very soon. 


> > ACCEPTING TRANSLATIONS: The Czech team accepts translations even from
> > translators who, as I say, just go along. All teams I've been involved
> > in do it. That's the spirit of FLOSS. I need to translate something and
> 
> There are several GNOME translation teams who work with similar rules
> about the assignments. I do not know details, but they also assigns
> modules to their translators.
> 
> > if it's done why not to provide it to someone else (to upstream in this
> > case). It does not work like this in the Slovak team. If someone
> > provided a translation he was told that he would become a long-term
> > translator or there was no interest... a very picky policy from the team
> > which has lost almost 40 % of translations in 10 last releases... and
> > result for users? no translations. There was a case when one guy
> > uploaded 60 % of Empathy translations. It's several hundred strings,
> 
> This guy and the emapthy is not example of such random translator. The
> translation was uploaded by regular member of our team. I think this
> translation was the first contribution of this translator.
> 
> > decent work. But he was told not to upload it until he gets over 90 %.
> > What happened? That guy got apparently frustrated and looks like the
> 
> That guy immediately reserved the module for translation. Yes. It is
> true that he submitted the complete translation after many months, but
> IIRC he did not copmlained when I asked for more complete translation.
> 
> There is no other intervention to this translation from me. The
> translation is still between the translator and reviewer (Peter).
> 
> http://l10n.gnome.org/vertimus/empathy/master/po/sk
> 
> > uploaded 60 % will never be submitted. The result for users? no
> > translations.
> 
> We really do not want to commit everything around. Sorry.
> Yeah, I have noticed. The problem is that you have not commited almost anything for years. That's why the drop from 81 % down to 45 %. As I said it's a very picky policy for the team which in such a bad situation.  


> > 
> > Because people kept asking why Slovak translations were in such a bad
> > condition I wrote an article about it on LinuxEXPRES.cz [1]. That
> > article got a huge response. And it's interesting readings, especially
> > commentaries by users, former and current team members etc. (the article
> > is in Czech and the commentaries either in Czech or Slovak). I received
> > e-mails from people and some of them suggested that another Slovak
> > localization should be made. If those ideas are brought up it's serious.
> 
> Yes. There are some people who disagree and complain and go around.
> There are other people who really want to understand and want to help.
> 
And they've got my full respect because in spite of everything we are talking here about
they still have will to do something with it. I'm not sure if I would be and I think most people are not, that's why they give up or run away to other Slovak team (e.g. KDE).

> 
> > I'm not a member of the Slovak team, so I can't take part in resolving
> > this problem. But please take it seriously. From my point of view, there
> > is no other real solution than changing the coordinator. I have followed
> > the development in the team for quite a long time. There have been
> 
> Thank you for following the development in our team. But I do not
> understand how you follow it when you are not a member of our team and
> you are not a member of the gnome-sk-list mailing list. You never
> contacted me, you never asked me, IIRC. Thank you.
> 
Do you know such a thing as a mailing list archive? You may be surprised, but they are available on the Internet. All messages in Vertimus are public as well. I have talked to many people from former, current members, members of other Slovak teams, and other people of the Slovak community. I have never wanted to make an intervention in your team. I thought it was a thing you had to resolve within your team. But I promised if they found out there was no way to resolve it within the team and had to go to public I would write what I have observed. As far as I know they initiated changes 5 months ago. You promised incremental changes and improvements, but the reality is that the situation is (surprisingly) about the same (well, all the process might be faster by 5 %). 

Because I don't want to send two responses I will briefly react to your second e-mail as well. You say there are 3 long-term translators and they support you. And you know what? I'm not surprised. One of my jobs is to look for new potential translators in Linux forums etc. In last two years, I have found several Slovak guys who wanted to give a hand. All of them either gave up immediatelly after they had found how it worked in the team or are pretty much inactive members. That's why I'm not surprised that the people who stayed in the team that long support you. Because those who disagreed are gone. As people who don't agree with the way the team works now will probably be gone sooner or later as well.
And please don't blame the number of translators in the team for the drop Slovak translations made in last years. As I said there is a reason why the team had just three members and even three members can do a lot if they are active and well-motivated and the process is fast(!). E.g. Czech translations of KDE are made basically by three guys and they are between 80 and 90 %.

Jiri