- From: Nathan Glenn <garfieldnate@gmail.com>
- Date: Tue, 1 Oct 2013 17:14:25 -0700
- To: Yves Savourel <ysavourel@enlaso.com>
- Cc: "public-i18n-its-ig@w3.org" <public-i18n-its-ig@w3.org>
- Message-ID: <CACs83pjh5DNuwJD-_rv_yjW-Nv3ELHjw_0EGywecfPn9B3JqbQ@mail.gmail.com>
That makes sense. I think it's good to keep anything ID-like off of the <source> element, too, since it might end up getting copied and renamed as <target>, which could lead to a duplicate ID. The XLIFF standard seems to agree, only allowing for an id value on <trans-unit>. For a part of WICS we wondered if we needed to parse documents invalidated by a duplicate xml:id value. Nathan On Tue, Oct 1, 2013 at 12:07 PM, Yves Savourel <ysavourel@enlaso.com> wrote: > > The other category that has the same source/ > > target confusion problem is idValue, which is > > currently mapped to resname on trans-unit. > > For that one I think maybe it's a case of non-mapping. > > We use idValue to link the content in the source document and the XLIFF > document. In the vast majority of the case, the source text is replaced by > the target text, so they have the same ID, and it make sense to use resname > for this in XLIFF. > > For the cases where the translation goes to a different node which has > also an ID, as in your case, maybe it make sense to say that the idValue > mapping applies only to source content ID because you are still likely to > be able to use targetPointer to merge back the translation at the right > node. > > The important part is to be able to identify the couple source/target by > an ID that is unlikely to change. I would say that even if the id of the > source is not the id of the target, it still provides a unique way of > identifying the pair. > > Cheers, > -yves > >
Received on Wednesday, 2 October 2013 00:14:53 UTC