Re: Explain-lite (Explain discussion)

> > I think that to re-invent explain in the context of its current model is
> > essentially a waste of time - the current model has been around for over 5
> > years now, it was seriously pushed by the ONE project, it was pushed
> > by the explain test bed and the number of implementations as far as I 
> > know is minimal. I dont see that changing the query syntax or
> > the ASN.1 definition of the record syntax is going to encourage a majority
> > of folk to think differently about implementing.

I do not completely agree with this. If there was a simple data structure
that mapped recommended named search points onto attribute lists (ala the
CCLInfo structure I posted) then you can write web clients and CCL clients
with only having to understand that one piece of Explain. As long as one
record has everything, then it should not be that daunting with documentation
directing people in the right direction (ie: you can use this one and forget
the rest for simple client applications). Z39.50 (currently) requires ASN.1
so developers have to support ASN.1 anyway. We created a separate record
syntax for CCLInfo (we did not expand the explain record syntax) so you
don't even have to show users the full explain record syntax.

Bob wrote:
> And XML is going to be different why??  TO me that is no different than a
> change to the ASN.1 definiton.

I am not pushing XML (don't care really), but one reason why XML is better
than ASN.1 is that it is simple to parse an XML document (without a DTD)
even if it contains constructs that are not understood. We use SNACC with
ASN.1, which means any ASN.1 construct added later will cause our decoder
to fail. This does not happen with XML. So XML is more extensible.

Alan

Received on Thursday, 16 November 2000 18:02:59 UTC