Our Translation Infrastructure team has some
concerns on the Section 2.6.7 "Target Content Modification".
This mail is about one of the the concerns.
Our main comments in blue, and questions in red. (For
non-HTML mail readers, I used "==>" to start a comment, and
brackets to mark a question).
The specification says about split/merge actions in the processing
requirements.
2.6.7.1 Without an Existing Target
- User agents may leave the existing target unchanged. ==> This is contradictory to the title.
No existing target.
- User agents may split the segment into
two segments.
- User agents may join the segment with the following one.
2.6.7.2 With an Existing Target
- User agents may join the segment with the following segment
- ==> No PR about splitting
segments. Does this mean it is not allowed? If <unit>
has one <segment>, then will it be prohibited that
agents split the segment? [We would like to prevent any split/merge
actions by agent. How?]
- User agents may delete the existing target and start over as
if working without an existing target ==>
If an existing target can be deleted, this implies the
processing requirements in 2.6.7.2 can be completely
ignored. [How can we
prevent it?]
Content in our translation kits is
pre-segmented and an agreed processing requirement is that
segments should not be altered (further split or merged). As it
stands there is no mechanism at unit level in XLIFF 2 that would
support our use case/requirement, and we would have to create a
proprietary extension to ensure our translation vendors would
support that PR, which implies losing xyz (standardisation,
interoperability, etc, you name it)
[Suggestions]
- Attributes (canJoin, canMerge) at <segment> level to
prevent changes in the number of segments inside a <unit>.
By default such changes can be allowed.
Or
- Alternatively a validation can be designed, but I think this
approach requires more sophisticated design.
Thanks
Oracle
|