migrated from Trac, where originally posted by clange on 20-Aug-2010 8:37pm
If we want to encourage publication of CDs in a machine-comprehensible way at their canonical URIs ("linked data", cf. my talk at http://cicm2010.cnam.fr/om/) – and the feedback about that hasn't been that bad so far –, some things in the spec need clarification.
- When introducing the notion of a CDBase and of CD and symbol URIs in chapter 2, it would make sense to add a sentence that a representation of that CD MAY (in the sense of RFC 2119) be made available for download at that URL. (Note that, of course, I don't want to enforce that!) – The abstract CD spec already says something like that: "![A CD base!] may or may not refer to an actual location from which ![the CD!] can be retrieved." From that place a back-pointer to the canonical URI format should be made.
- When talking about CDURL and about CDGroups referencing CDs, it would make sense to say that in the absence of a CDURL, an application can try to dereference the canonical URI of a CD in order to download it. (BTW, why do we need CDURL anyway? What sense does it make to tell someone who already has the CD where to find it? Is the use case a developer or system that has a local, maybe outdated copy of a CD and wants to know where to find the official reference implementation of that CD?)
- CDGroupURL needs a similar treatment.
Note that I'm pretty much open to discussing in what formats CDs should be made available. HTTP content negotiation supports offering different formats from the same URL. Maybe we should suggest a few reasonable ones. I could imagine:
application/openmath-cd+xml → OCD XMLapplication/openmath+xml → The CD as OpenMath object encoded using the meta CD (note, that is a new idea – but probably the most reasonable one)application/rdf+xml → RDF metadata of the CDapplication/xhtml+xml or text/html → rendering of the CD (what you currently get when you dereference e.g. http://www.openmath.org/cd/arith1)
migrated from Trac, where originally posted by clange on 20-Aug-2010 8:37pm
If we want to encourage publication of CDs in a machine-comprehensible way at their canonical URIs ("linked data", cf. my talk at http://cicm2010.cnam.fr/om/) – and the feedback about that hasn't been that bad so far –, some things in the spec need clarification.
Note that I'm pretty much open to discussing in what formats CDs should be made available. HTTP content negotiation supports offering different formats from the same URL. Maybe we should suggest a few reasonable ones. I could imagine:
application/openmath-cd+xml→ OCD XMLapplication/openmath+xml→ The CD as OpenMath object encoded using themetaCD (note, that is a new idea – but probably the most reasonable one)application/rdf+xml→ RDF metadata of the CDapplication/xhtml+xmlortext/html→ rendering of the CD (what you currently get when you dereference e.g. http://www.openmath.org/cd/arith1)