Re: PO-based Documentation Translation

[Danilo Segan <dsegan gmx net>]

Hi Dave

субота, 27. септембар 2003. 22:19:05 CEST — Dave Malcolm написа:
> I've been lurking in this discussion and wanted to ask: is there
> anything we can do within Conglomerate to make it easier to translate
> DocBook documents?  (Conglomerate aims to be a user-friendly XML
> editor for DocBook and other XML formats, for those who don't know)

While I consider Conglomerate to be a really nice XML editor (or at  
least that it's going to be ;-), I don't think it's up to the task for  
other things than the *initial* translation (like any other XML editor  
tool).

The problem in here is that PO files are generally considered *sets* of  
elements, and their order is irrelevant (from technical point of view,  
but they're usually ordered in some way in PO files, and that does help  
translators, because they know that only related stuff are near in the  
file).

OTOH, XML files, and particularly DocBook documents, are structured.  
That, by itself, means that transformation from one to the other is not  
really simple.

The key thing in here, that concerns us translators, is how to make  
these two cooperate. As a translator, I care to see only the text to  
translate, and not the DocBook markup.

So, a tool that we need would transform a document of the form:
<article>
<chapter>
<title>First chapter</title>
<para>This paragraph introduces us to FOO.</para>
...
</chapter>
</article>

(these are not correct tags, it's just an illustration ;-)

And translators care only about:
msgid "First chapter"
msgid "This paragraph introduces us to FOO."

Even if you later add a new chapter in front, which begins similarly to  
the existing one:
<title>Second chapter</title>
<para>This paragraph introduces us to BAR.</para>

Tools that already exist (msgmerge) would *automatically* translate  
most of the new paragraph, because "fuzzy matching" works great for  
strings as similar as the two paragraphs in my example.

Reordering paragraphs, small changes (eg. adding a "." at the end of  
chapter names) are hard to track in any structure-oriented editor, so  
that's where PO tools have advantages.

Cheers,
Danilo