Notes from the conference call

["Yves Savourel" <ysavourel@translate.com>]

No enough people were there for an official meeting but we
had some discussions.
Here are the notes.

Cheers,
-yves

Segmentation Subcommittee Meeting
October-12-2004

Present: Magnus, Tony, Yves

Not enough for a corum, but we had some discussion.
Here are a few notes from it:


We looked at Magnus examples cases posted recently and tried to apply the two main solutions that have discussed:
- using some inline elements (i.e. <mrk>)
- using some grouping of trans-unit

- Discussed the case with <g> elements that needs to be segmentated.
The solutions are (for <mrk>)
- using empty <mrk/>, but this is juts moving the problem out of XLIFF to whatever tool will read it and work at the segment level.
- adding new closing/opening (but need to know whether you can do this, so may need a mechansim)
- forbid segmentation


- Discussed <alt-trans> case, where <alt-trans> corresponds to part of the source only.
-> solution would be to add a segid in the alt-trans
works ok with <mrk> approach

For re-segmentation approach:
not as good because you don't have an alt-trans at the group level (if group encloses segmented <trans-unit> but there is only a match for the original item (unsegmented).
one possibility is to replicate the <alt-trans> at each <trans-unit>
another solution would to promote <alt-trans> at the group level. But that's not very clean.


For tying together the segments in source and target we need some kind of segid
unique within a trans-unit would be enough.
any XSD type would allow this kind of scope?


end of discussion