[xliff] Updated Specifications

[Yves Savourel <ysavourel@translate.com>]

Hi everyone,

The specification and the schema of the latest draft update are attached.
here is also a few notes for your next lunch-break meditations:

-- The reformat attribute:
Text and XSD entries for reformat have been updated. The values of reformat
cannot be easily validated now. I suppose we could come up with a regular
expression pattern... But it would be a complicated one (Bryan,
Shigemichi... if you feel like it...)

-- One more reversal needed:
There is one more change we voted for 1.1 that needs to be reversed for
backward compatibility: Forcing at least one <group>, <trans-unit> or
<bin-unit> inside a <group> (in 1.0 the <group> element can be empty).

-- The renegade Chinese character:
It was an 'é' (in 'Gérard') and has been fixed. Probably some incorrect save
into Latin-1 at some point.

-- The XLIFF tree structure
I would propose to drastically modify Appendix A (XLIFF tree structure) In
my opinion the full tree doesn't really add any information, and now that we
are back to un-ordered elements in many contents, it'll be just very messy
and almost unreadable. And it's difficult to maintain. Instead I propose to
use partial diagrams of some of the main elements to give the reader a
better overview.

-- The purpose attribute:
I've noticed a problem with the new attribute 'purpose': It's currently a
list of pre-defined values with possibility of extending it. And at the same
time, it says several values can be used together in a list (just like
reformat). The way our schema is defined currently such list would be
invalid since it's an enumeration. I'm also not sure about extended values:
this is not the same type of value as a normal list-attribute. We could
probably come up with a pattern that would validate the attribute, but like
for reformat it won't be a simple one.

-- XML Schema:
I've included a XLIFF document called Everything.xlf in the Zip file. At
some point it will have all elements and all attributes possible. I've
started from a sample file generated by XMLSpy and added a lot manually.
While doing this I had the nasty surprise of discovering that our current
schema doesn't quite work yet. For example, I'm having hard time figuring
out how to write the rules for <source> (currently there is problem with
mixed content). If anyone has some time to help out, I'd appreciate.


kenavo,
-yves

Attachment: XLIFF-1-1-Draft-3.zip
Description: Zip compressed data