Comments on: …and here we go again

By: choeger

choeger — Wed, 17 Dec 2008 12:35:54 +0000

Yepp, pure community translation won’t work. But I think that any .po maintainer is pretty bored by translating the same sentences again and again. That leads to slow progress I have my own .po patches waiting for review on g-b-o since months.
A single DB where a lot of suggestions are stored could really speed up things (and help ensure quality also, when the transifex step is done by pros)

By: Runa

Runa — Tue, 16 Dec 2008 16:51:48 +0000

*nod*

Thats the best approach because besides giving an early start, it effectively also allows corrections in the English strings (lots of them are spotted by the localizers) to go through fast. During string freeze, developers generally have to wait for permission from the localization group to make changes.
But yeah.. localizers also have to be mentally prepared to rework on bits because things change really fast.

By: mui

mui — Tue, 16 Dec 2008 16:25:27 +0000

Sorry, I missed this sentence “and then be reviewed by a professional .po maintainer.”

You are right, a real pro is crucial.

By: mui

mui — Tue, 16 Dec 2008 16:20:18 +0000

@choeger: I’ll tell you one story about our (classic commercial) product. We had translations done by unprofessional translators – native speakers. Then one day we started to get really weird support calls. It turned out, that one translator messed up one of the strings. It wasn’t a huge mistake but it was enough to cause confusion among users. It took us 6 months to pinpoint the issue to the incorrect translation.

Do not underestimate l10n.

By: Kevin Kofler

Kevin Kofler — Tue, 16 Dec 2008 15:57:16 +0000

> In fact with “string freeze” (or, the time when the developers hand off the English versions to the translators) the effective window to actually translate is around 1 month.

Solution: Actually start translating as the software is developed, don’t wait for the string freeze! Then you have more time.

By: choeger

choeger — Tue, 16 Dec 2008 11:24:08 +0000

Translation _should_ be for hobbyists. Pretty all people that could deal with a given translation system, read the policy and so on, will speak (or at least read) English fluently. That means, they have little to no interest in translating things.
That has tow consequences:
1. Translations need to be really easy (on a “translate that string”-base)
2. Translations should re-use all translated strings they can get

That implies: One should set up a String-translation database to which normal users could send single fragments just when they come to their attention, po files (and other stuff) should be filled with the content of that db and then be reviewed by a professional .po maintainer.

How about that?