Re: Shared references service

Hi David,

On Jun 22, 2012, at 15:31 , David Carlisle wrote:
> For those of us using xmlspec, I don't suppose you could offer xml as
> well as json? It doesn't make it much easier but it might make it a bit
> easier: the json parsing in the following is a bit flaky and it would
> probably be easier for the server just to serve the structure it had?

If it means more people using the service, and more people contributing to maintaining the DB, then I certainly could!

If I did, would you want me to use the same syntax you exemplified? I can think of simpler ones, but I don't know what your exact needs are.

One potential problem is that a non-negligible number of the references don't (yet) have the nice title/href/authors/publisher split fields but rather just have an html field that contains the pre-rendered reference (they're being phased out, but I still see 890 such references versus 5491 in the new style). There is no guarantee that those are well-formed for XML consumption (most probably are, but it's not guaranteed). So the safe thing to do there would be to return those in a CDATA section — but I don't know how helpful that would be to you. The alternative is to not return those at all.

> If not, the json parsing could of course be improved, this is just a
> quick test of your server:).

Yeah, in general I have to say that I don't think it's very practical for XSL/XQuery not to have native JSON support ;-)

> I note that q=MATHML returns the latest MathML draft in TR. I don't really mind but the notes imply it perhaps ought to return the editors' draft (or there should be another ID that returns the editors' draft?)
> which would be
> 
> https://www.w3.org/Math/Group/draft-spec/
> rather than
> http://www.w3.org/TR/MathML/

As indicated in the notes (but perhaps not clearly enough) the idea is that what should be returned ought to be the "latest and greatest". The exact interpretation of that is entirely left up to whoever inserts that reference. Sometimes the ED is far ahead and better maintained than the TR version, sometimes the ED can be a big mess and there are frequent cleaner snapshots in TR.

If you think that the ED makes more sense for MathML, I strongly encourage you to give me your GitHub ID so that I can add you as a contributor and you can make the change ;-)

(If you're really lazy, I'll do it though.)

-- 
Robin Berjon - http://berjon.com/ - @robinberjon

Received on Friday, 22 June 2012 14:01:35 UTC