The article starts with an excellent display of why a "Key = Value" solution is not well fit for localization and the subtleties of different major languages.

But then vaguely explains the solution to the problem and fails to address more complex cases, which are very common (like those where multiple amounts, custom names and genders may be placed in the same sentence).

XLIFF may be standardized but the signal to noise ratio is horrible, and no tools to translators were mentioned.

Then proceeds to proceeds to give links to paid localization solutions without still explaining how the problem can be solved efficiently.

This is why I reviewed the article as incomplete.

Thanks for the feedbacks. Will add more solution related information in the article.

Putting article itself aside, Russian plurals for apples aren't correct.:

0 ?????

1 ??????

2 ??????

...

5 ?????

...

etc.

Thanks for letting us know Zaoshi! I've corrected the mistakes. I have included the form for 10.0 as "other" in the example.

Interesting article, good to point out some of the differences. The latter half is a bit short IMHO. It seems there would be different approaches whether you need to translate full text or assemble text from several pieces.

This however:

nplurals=3; plural=(n%10==1 && n%100!=11 ? 0 : n%10>=2 && n%10<=4 && (n%100<10 || n%100>=20) ? 1 : 2);

is Daily-WTF-material, that's obfuscation :)

For what it's worth, I put together a simple tool to be able to use Gettext in my Android game, as I wanted proper plural support. You can find it here: https://github.com/agateau/linguaj. The nice thing about Gettext is that there are plenty of tools to use it, so it's reasonably easy to get someone up to speed. A cross-platform tool like Poedit, for example, is simple to use even for someone who is not a developer.

The problem with Gettext is, keys are English phrases. If you so much as add a comma, you have to search-and-replace everywhere that phrase is used. Inconsistent and error-prone.

Alternative #1: Use a simple key-value translation table. Non-techie translators can understand that just as well as PO files, and they don't need special tools, which are all overpriced garbage anyway. Translators HATE HATE HATE having to use $1000 crapware when Notepad would do just fine.

Alternative #2: Use vagueness and symbolism to sidestep proper pluralization/gender altogether... because it adds approximately nothing to gameplay and takes time+money that could be spent making the game enjoyable. It tends to come out amateurish anyway. Usually the English isn't even right, and it's safe to say the Chinese translation is inverse Chinglish. It takes a big budget to do it convincingly.

Ummm... end rant :)

I don't understand what you mean with "inconsistent and error-prone": it's up to the translators to update their translations once you give them the English text. Ah, maybe you mean you want to be able to update the English text without invalidating the translations? A workaround for this is to use "developer english" and "real english", but that's a bit cumbersome.

Regarding alternative #1: key-value translation tables do not handle plurals, and many tools which work on PO files are free software, they may be garbage, but they certainly aren't overpriced :)

I agree with alternative #2: especially since it makes it possible to build games for kids who can't read yet. It's not always possible nevertheless, for example describing achievements without text sounds difficult.

The Daily-WTF rule is extracted from PO as an example of defining the logic for different languages. You can check out the rules from the unicode site I mentioned in the article as well.

And yes PO is another flexible enough format to work on the pluralization. The English key problem could be solved by using some standard keys instead.

Ok, @Aurélien and @onesky, "developer english" / "standard keys" seems like the best practice if you're using PO/gettext/etc. Still, if I were writing code like renderText(__("button.menu.about_us")) then I could just as easily write renderText(TR.button.menu.about_us) using the programming language's regular facilities. No additional tools or file formats needed - a key consideration for solo developers and small teams.

Hi, guys. To any of you who is in need of a good collaborative translation tool that supports plurals, I warmly recommend the localization management platform https://poeditor.com/

It's reliable and the support team is very swift.