> Date: Fri, 28 Nov 2003 09:47:35 +0000
> From: Robert Sanderson <[log in to unmask]>
>
> > modifier = '/' modifierName [ '=' modifierParam ]
>
> I still disagree that this is necessary, as then our current
> modifiers will look like:
>
> any/stem=true/phonetic=true/relevant=true...
>
> If the reason is to make the queries more readable, then I think
> we're failing as PROX is by far the least commonly used bit of CQL
> compared to modifiers in general.
As you'll have seen in Alan's grammar, the proposal is for the
``comparitorSymbol modifierParam'' part to be optional. I think
that's cool.
> Please, in favour of resolving 1.1 blocking issues, can we leave
> this discussion till 1.2 ?
I'd much rather sort this out now, one way or another. We really
don't want to be making a fundamental change to the CQL syntax in
version 1.2.
> > Reserved words are 'and', 'or', 'not', and 'prox' (case
> > sensitive). When a reserved word is used in term, the term value
> > is the word.
>
> CQL is and should be case insenstive.
It's case-insensitive now, yes. I have no strong feelings over
whether or not that's how it should be, but the Principle Of Least
Buggering About suggests we should let sleeping fruitbats lie.
> > The values of index, namedComparitor, and modifierName are
> > structured and must be of the form [prefix.]name where the prefix
> > defines the context of the name. If omitted, the default context
> > is used.
>
> For indexes only. The default context for all others is cql.
Correct.
> > The default context for queries (unless redefined in the
> > scopedSearchClause construct using '>uri') is the reserved context
> > name 'cql'.
>
> The default context for indexes is server defined. For SRW/U, it's
> //explain/configInfo/default/@type=contextSet in ZeeRex
Also correct.
> > weight=<integer>
> > Associate a normalised weight with this term in the query
> > (0..1000). (From the new Attribute Architecture.)
>
> Does anyone use this? If so, for what?
That's a profiling issue. I'd argue that most of the things Alan's
tried to stuff into the CQL context set fall into this category, too:
we don't need to standardise them as part of what CQL is, so let's
not.
_/|_ _______________________________________________________________
/o ) \/ Mike Taylor <[log in to unmask]> http://www.miketaylor.org.uk
)_v__/\ "South Yorkshire mass murderer - how can you sleep at
night?" -- Manic Street Preachers.
--
Listen to my wife's new CD of kids' music, _Child's Play_, at
http://www.pipedreaming.org.uk/childsplay/
|