RE: [search-ws] [Issue] XCQL: do we need it?

["LeVan,Ralph" <levan@oclc.org>]

If it wasn't clear that we do NOT accept XCQL queries, then that is an
oversight.  We originally did support that in SRW (SOAP messages) but
saw absolutely no value in it and dropped it.  It has NEVER been
supported in SRU requests.

Ralph

-----Original Message-----
From: Raj Singh [mailto:rsingh@opengeospatial.org] 
Sent: Wednesday, October 17, 2007 4:54 PM
To: LeVan,Ralph
Cc: Farrukh Najmi; search-ws@lists.oasis-open.org
Subject: Re: [search-ws] [Issue] XCQL: do we need it?

Let me second Ralph's point (but also agree with Farrukh!). In one of  
my projects we've struggled with the need to round-trip queries with  
UIs to edit them. The structure of an XML-based query format makes  
this a reasonable process, in comparison to parsing raw CQL (or SQL).  
On the other hand, I agree XML is a pain when it comes to  
serialization and messaging.

I'm not sure what the right answer is here. One thing I've noted in  
the years I've been working on my project is the lack of really good  
SQL editing UIs. If that industry hasn't solved the problem with  
decades to work on it, will we?

The only solution I can come up with is that servers should be able  
to output the XML format--which solves the client UI issue--but  
servers only receive requests in the non-XML format--solving the HTTP  
request transport issue. This puts the main burden of complex query  
parsing on the server. This shouldn't introduce extra  
interoperability issues because servers probably have to do this  
anyway to translate into their native query language. For clients  
it's probably trivial to translate from the XML format to raw CQL.  
What do you think of this proposal?
---
Raj


On Oct 17, 2007, at 4:05 PM, LeVan,Ralph wrote:

> Here's the reason why we have it and why we'll want to keep it.  If  
> you
> have a better way to accomplish this, I'd be happy to hear it.
>
> The problem is that many of us use thin (browser-based) clients to our
> SRU servers.  They have no ability to parse the query when it is
> returned to them.  If we want to be able to use parts of the query  
> in a
> display or allow the user to modify the query in any way, we need a
> parsed version of the query.  Hence the XCQL in the
> echoedSearchRetrieveRequest element.
>
> Ralph
>
> -----Original Message-----
> From: Farrukh Najmi [mailto:farrukh@wellfleetsoftware.com]
> Sent: Wednesday, October 17, 2007 3:35 PM
> To: search-ws@lists.oasis-open.org
> Subject: [search-ws] [Issue] XCQL: do we need it?
>
>
> The strawman spec refers to XCQL (btw citation is missing). XCQL  
> sounds
> much Filter syntax within RegRep 3.0 and OGC Catalog specs.  It  
> provides
>
> an XML representation of the CQL syntax.
>
> Based on my experience such filter syntax has not been particularly
> useful as it is very verbose and complex. Also, it typically requires
> HTTP POST rather than the more desirable HTTP GET.
>
> I suggest we simplify the spec and drop any reference to XCQL.
>
> -- 
> Regards,
> Farrukh
>
> Web: http://www.wellfleetsoftware.com
>
>
>