Re: Do not translate <placeholder-X/> tags (xml2po)

[Chusslove Illich]

> [: Bruno Haible :]
> I can basically imagine three kinds of "must not translate" constraints
> relating to XML: [...] Would that be something useful?

Given experience so far, I have doubts.

>   - xml-markup         preserve the fact that it's XML, but free to remove,
>                        add, or replace XML elements.

While this would mean that the translation is checked to be well-formed, it
would pass through otherwise mistyped tags/attributes for the given XML
specialization. For example, in my favorite editor I press a shortcut, enter
foo, and get <foo></foo> out; so if I mistype the tag, the XML is still
well-formed.

Also the question of how it would apply to HTML markup (if at all), as
opposed to XHTML.

>   - xml-structure      preserve the XML tags, but free to change the text
>                        inside and outside the XML elements.

This would mean that translation must follow the original by tag set. For
example, when XHTML is translated, CJK translators tend to remove bold tags,
as bold ideographs are hard to read in text body sizes. Also, a translator
well aware of given markup may want to improve it compared to the original
(e.g. add a Docbook tag around a segment where the original does not have
it).

>   - flat-xml-structure preserve the XML elements and their contents, but
>                        free to change text outside all XML elements.

Same problem as previous, plus making the author, rather than translator,
decide what is to be translated and what not. Even literals that the
"machine" should understand are within doubt: translator may want to add a
short note, or surrounding formatting (whitespace, quotes) within the tag.
And I certainly wouldn't want to periodically enter into discussions if a
proper name or another should or should not be translated...

In conclusion, while I consider the proposed xml-format good to have (it
would catch, in my estimate, >95% of usual markup errors[1]), the other two
I'd flat-out recommend against.

[1] There's a sample of XML errors at
http://www.kde.cat/aacid/xml_errors/trunk/, where 'Messages' column is for
POs containing XHTML and one other XML, and 'Docmessages' for Docbook POs;
among all the errors, only those complaining about unknown tags and
attributes would not be caught by xml-format check.

-- 
Chusslove Illich (Часлав Илић)

signature.asc
Description: This is a digitally signed message part.