Hi Rodolfo, Josep, Pål, all,
>> That's because, like Rodolfo interprets that the content of <target>
>> can be blank, one can also interpret that a copy of the source is
>> simply the translation in its initial form in a many steps process.
>> After all the specification doesn't define what "translation" is.
>
> Well, actually my tools don't put <target> elements in untranslated
> segments, not even blank ones. So, not all tools create invalid XLIFF
> files.
Actually they do. I can use the latest version of Swordfish (3.0-11 7DC-3-9) to save an XLIFF file that has both blank <target> elements and <target> elements with a copy of the source.
The bottom line is that one or more attributes are the only way to indicate what exactly is in <target>.
In 1.2 none of those attributes were required nor had default values, which make many valid 1.2 files difficult to interpret.
We simply need to come up with a good set of attributes/values and processing expectations in 2.0 to reflect the different states of the segments' translation.
Cheers,
-yves
