simple search api (was Re: mimetype standardisation by testsets)

[Fabrice Colin]

On 11/28/06, Mikkel Kamstrup Erlandsen <mikkel.kamstrup at gmail.com> wrote:
> 2006/11/27, Joe Shaw <joeshaw at novell.com>:
> Wouldn't returning the full metadata (except snippets) for every single hit
> be a costly affair? Maybe we need a parameter for which fields to return.
>
Agreed. Who knows how many metadata fields there might be...

> > We use a separate API for text fragments in Beagle because extracting
> > them is extremely slow.  I would recommend against it in the query API.
>
> If the queryDetailed method took a list of metadata fields as input
> "snippet" could be one of them. I still think we should have a separate
> method for it though.
>
The fields list as input sounds good to me.
As for how slow obtaining the snippet is, this depends on the
implementation, eg generate from the full document text versus from
the terms positions list, but having a separate method would be
useful, for instance to a client that supports hiding/showing the
snippets field.

Fabrice