Re: Gnome2 User Guide: translate it!

[danilo gnome org (Danilo Åegan)]

Hi Clytie,

Today at 8:51, Clytie Siddall wrote:

> On 23/02/2006, at 5:09 AM, Danilo Åegan wrote:
>
>> By the mighty Shaun, our fellow GDP leader, we present you Gnome2 User
>> Guide in translatable PO files:
>>
>>   http://kvota.net/doc-l10n/by-modules.html#gnome2-user-guide
>
> This is a great move! :D

I only changed it to gnome-user-guide instead of gnome*2* (to follow
the package name) now.

> I don't know if the GUG is the same, but when I was trying to
> translate the Debian-Installer User Guide, I found the length of many
> of the strings intimidating and frustrating. PO format doesn't work
> very well for l-o-n-g, complex paragraphs.

That's why xml2po tries to hard to split them wherever possible.
Debian is not using xml2po for their own purposes, and I don't know
what is their procedure.

> Firstly, you struggle away for ages and end up with ... 3 strings
> completed.
>
> Secondly, it's harder to find errors in l-o-n-g, complex paragraph
> strings.

Now, our stats generation is using a different PO merging script from
msgmerge to provide diffs which help greatly in such situations (small
changes in complex paragraphs).

I've even talked about it on last GUADEC in Stuttgart, and even
blogged about it here:

  http://danilo.segan.org/blog/prevod/po-merging-and-diffs

So, simple and small fixes should be easier to track and correct in
translations if you grab PO files from our status pages.

> I know we need cogent strings, but I'd be strongly in favour of
> splitting long or complex paragraph strings into sentence-strings, or
> at least shorter strings, if that worked for the text.

I've been thinking about it, and it's not too hard to implement, BUT,
I'd rather have sentence-splitting in a separate tool: you sometimes
need to re-order sentences in the translation and it wouldn't be
possible then.

> Linguistically, it should work, since a sentence is a complete
> statement of its own, but that will depend on the text. ;)

You sometimes also want to lose a couple of sentences, if something
can be more clearly said in a translation.  I figured the
paragraph-level was the right choice that would work for everyone.

What I can consider adding is "Sentence-Level: 0/1\n" PO header field,
and deciding on that what separation level we want.  But, this
complicates things a bit and I won't be able to work on that anytime
soon.

> I must admit, I struggled through three or four of the _shorter_ DIUG
> files ... and haven't been back. The longer files just intimidate me.
> If their strings were broken up (correctly) more, I would take them
> on.

Well, the idea is to compare translation through PO files with direct
translation in XML files: we are going a step at a time :)

Cheers,
Danilo