The ways of the translators [was Re: (no subject)]

Ok, I'm starting a new thread on just the gnome-i18n, since it's the 
only one where I expect interest (sorry if I am mistaken).

Glynn Foster wrote:

>I seem to remember some of the Sun people saying that translations were
>pretty inefficient in GNOME - maybe it's time to rethink? [1]
Here, I'll describe a way we at the Serbian translation team are doing 
the translation. I think we're not too fast, but are fast enough (in 3 
weeks we got to around 20000 officialy translated strings, over 22500 
unofficialy [some unfinished translations, like 25% of Evolution, 
etc.]), and I put it all on the model. Just for the record, I was the 
"most active hacker" in Gnome CVS in latest summary without touching a 
line of code :-) And even Christian Rose stole some of our fame by 
commiting some 10-15 translations :-P

We have a website designed for this particular task. It reminds a lot (a 
real lot, actually modeled after)[1] pages, with 
some interesting additions. A "registered" translator can choose a 
translation to work on with no obstacles, (s)he just has to point at a 
button next to the program needing a translation and/or update that says 
"Take the job". At that time, that translation is "taken", and noone 
else can interfere in the process, but of course, when the translation 
is well and done, any other translator can either update it, or just 
report a bug on it. The "statistics" pages are updated automatically.

When someone takes over a "untranslated" program and sends a translation 
for it, (s)he is given the "maintainer" status. It's changeable after, 
but they claim the responsibility now. Whenever someone else updates the 
translation, it awaits the confirmation of its maintainer. (Firstly, 
this was all delegated to the "administrator" of the translation, but it 
appeared to be too closed compared to the other features of the model.)

Also, the site contains other "support tools" for translators, like to 
completely open (even to anonymous users) dictionary with voting 
mechanism (to resolve disputes), with one member writing a real simple 
Gtk+ dictionary program which reads the database of the site dictionary 
directly (perhaps we should have investigated using the GtkDict :-)), 
some tutorials and manuals, etc.

The main feature of the concept is it's openness. It seems that many are 
discouraged to involve if they must engage in perpetual discussions 
instead of doing the real work. So, with a chance to just register, and 
do the job, there were many who actually decided to register. Only some 
of them actually contributed (of nearly 40 registered members, just 15 
actually contributed, but it's not that small a number for a number of 
translators, especially "active" translators), but some of them 
contributed to the dictionary, and other things. I guess this approach 
"I don't need to get approvement from anyone" was what attracted 
translators. By the way, a point system based on translations someone 
engaged in was also introduced, but it's just a means of encouraging 
translators who enjoy that kind of thing.

Also, the coordination provided by the site, which avoids duplicated 
efforts, helps track the current status, avoids dead-ends (someone 
taking the responsibility for the translation, and later disappearing is 
regulated with automatic deadlines), is a substantial contributor to the 
(I think) overall success.

There are many details I never even touched, but I hope this will prove 
to be an interesting way of doing the translation, even though it has 
some flaws (inconsistency being the first and foremost, but we're 
planning on fixing it in pre-release states and when we decide on the 
words in dictionary; if some of that is agreed upon prior to 
translation, that much better).
The other big flaw is synchronization with Gnome CVS, which must be 
performed manually (though, it can be at least a bit automated through 
the d.g.o/prj/gtp/status).

For anyone interested, feel free to skim over, 
but guess you won't understand it unless you can read cyrillic and are 
at least familiar with the artificial language "Slovio" (though, it's 
based on latin script, I guess it can help :-)

I'd like to hear any comments on the idea, potential flaws, and maybe 
follow-ups. Does Gnome have the resources to provide similar framework 
for all translation teams? I'd be willing to work on such a software 
with other translators and developers. Of course, some teams would 
require more strict control, but that's easily achievable, and not a 
real problem (just revert to our first approach of administrator-approval).

>[1] Okay, okay, that was probably a bit too controversial for this
>thread ;)
So, here's the new thread for just *those* issues. :-)


[1] Yes, it's a famous page

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]