Re: More on xml errors in documentation

[Daniel Mustieles García <daniel mustieles gmail com>]

So, we could use the strict check as the default one, just keeping in mind it can shows some (rare) false positives, right?

The other question about this is, is it possible to do this check in DL to show this typos (like the errors generating POT file are shown, for example)? It would be realy useful to have this check, so translators and coordinators would be aware of them.

Cheers

2013/4/8 Shaun McCance <shaunm gnome org>

On Mon, 2013-04-08 at 11:30 +0200, Daniel Mustieles García wrote:
> 2013/4/6 Gabor Kelemen <kelemeng gnome hu>
>
>
>         I think these kinds of warnings are useful, until it is not
>         compulsory
>         to fix all of them :).
>
> Why not? I'm not sure about it. If Documentation team adds some tags
> to the original strings, translators should keep those strings. Also,
> having freedom to add our own tags to translated string may break the
> building process so fixing those errors should be fixed.
>
> I don't know how Mallard documents work, and how translators have to
> deal with tags (I mean: should we add or remove tags when translating
> strings? Shouldn't it be done by the Documentation team?) I think
> translators shouldn't change the format adding or removing tags
> (althouhg it doesn't break the compilation), but having some comments
> from docs team would be really useful.

I think there are cases where it's worthwhile for translators to add
tags, but they're probably the exception. Thinking out loud:

* East Asian languages might want to add Ruby markup. And this will
actually work correctly in 3.10.

* Some languages might want to use an English word for a technical
term when there isn't yet an established local word. They might want
to mark this with <em>.

* The instructions for certain actions on different keyboard layouts
might involve more or fewer <key> elements than for en-US.

If we really end up getting a lot of false positives, or if we want
to turn these checks into hard pre-commit checks, we could add some
sort of markup to allow translators to say "I know what I'm doing in
this case". But that can be a hassle, so it's only worthwhile if we
have a real problem with false positives.

--
Shaun