Closed jayvdb closed 8 years ago
Yes. According to my reading of the RELAX NG spec, these are equivalent, see section 4.8 here:
https://www.oasis-open.org/committees/relax-ng/spec-20011203.html#IDAEAZR
Due to the way the schema generation is done, doing it this way add less complexity in rnc2rng, and should be more optimal during validation because the simplication step doesn't have to be done.
Does this cause problems for you? If so, why?
The empty ns in <name ns="">class</name>
is my main concern.
I havent yet tried to use it. Will do so over the weekend.
It looks like it does cause a problem. https://travis-ci.org/jayvdb/citeproc-py/builds/97387207 (I'm still digging into the cause..)
Hmm, that thing is not very explicit on what is failing though.
This issue was not the problem causing the tests to fail. The citeproc schema is fixed with #6.
fwiw, I started to 'fix' this, and have a partial solution at https://github.com/djc/rnc2rng/compare/master...jayvdb:issue_4 I can continue it if you like.
I'd prefer to keep it as-is, for the reasons outlined above.
When I do a diff between trang and rnc2rng results for https://github.com/citation-style-language/schema/blob/master/csl.rnc (i.e. this is a continuation of #1), the only difference is
name
is being moved from an attribute to a child node 'name'.That type of change occurs hundreds of times in the output.