Re: Query related comments (was RIM v0.54 distribution)

[Farrukh Najmi <Farrukh.Najmi@east.sun.com>]

David,


At the moment, I (and the query experts I have talked to) believe  the approach
that you  have been advocating (also the NIST approach) as tantamount to
rolling your own query syntax. Just because it is XML does not make it anything
else. It is not based on any standard. It is not familiar to clients. And most
of all it is difficult to spec.

I assume that you will demonstrate how RIM based sample queries and their
combinations will be represented in the approach you advocate soon. We need
something analogous to the queries I published in both XPATH and OQL syntax.

Lastly, we need a mind meld as a team on this issue soon since time is running
out.

--
Regards,
Farrukh



David RR Webber wrote:

> Message text written by Farrukh Najmi
> >BTW I have had significant discussions with several query language experts
> (people who have contributed to existing query language standards) and
> their
> general opinion is that designing a home brewed query interface syntax is a
> very difficult task to get right. They unanimously favor a constrained
> subset
> of an existing standard. These experts include Rick Cattell (OQL), Umit
> Yalcinap (EJB-QL), Neelam Vaidya (XML Query) and Eve Maler (XPointer).
> <FullDisclosure>They are all Sun employees</FullDisclosure>.
>
> Lastly, I want to propose a requirement that any query interface we come up
> with must be able to do combinatorial queries using all attributes defined
> by
> RIM. Note that this is a much simpler requirement than designing a general
> purpose ad hoc query mechanism where there is no pre-defined information
> model
> or schema.<
>
> >>>>>>>>>>>>>>>>>>>
>
> Farrukh,
>
> The proposal I documented today appears to meet both these criteria
> completely.
>
> Since it is only a XML container for business functional procedures and
> focused subset queries based on a constrained subset of XPath/XPointer,
> and PLUS it is exactly tied to exploiting and providing combination queries
> that exactly reflect the RIM access paths.
>
> Most especially because it completely avoids at the procedure level
> of dictating ANY query language whatsoever - just the business functional
> API behaviour at a high level.
>
> I can work tomorrow on producing a quick two or three page proposal
> to formally qualify this beyond the details in the emails as indicated
> by Scott previously prior to Friday.  The email does however appear to
> suscintly capture the salient details.
>
> Thanks, DW.


begin:vcard 
n:Najmi;Farrukh
tel;work:781-442-0703
x-mozilla-html:FALSE
url:www.sun.com
org:Sun Microsystems;Java Software
adr:;;1 Network Dr. MS BUR02-302;Burlington;MA;01803-0902;USA
version:2.1
email;internet:najmi@east.sun.com
fn:Farrukh Najmi
end:vcard