Re: [MORG] Review: draft-ietf-morg-fuzzy-search-02

[Timo Sirainen <[e-mail]>]

On Tue, 2010-11-16 at 16:20 +0000, Dave Cridland wrote:
> > > > Yes, I think SORT (RELEVANCY) should be required for that.  RFC  
> > 5267
> > > > isn't super clear on if SORT supports PARTIAL, but I guess it  
> > should
> > > > according to ABNF.
> That's basically what CONTEXT=SORT is for:
> 
>    Servers advertising CONTEXT=SORT also advertise the SORT  
> capability,
>    as described in [SORT], support the extended SORT command syntax
>    described in Section 3, and accept three additional return options
>    for this extended SORT.
> 
>    These additional return options allow for notifications of changes  
> to
>    the results of SEARCH or SORT commands, and also allow for access  
> to
>    partial results.
> 
> I don't think that's anything less than super-clear.

Oh, right. I was mostly just reading through sections 3.* and 4.4.

> > I was planning on using this example:
> > 
> > C: D02 SORT RETURN (PARTIAL 1:10) (RELEVANCY) UTF-8 FUZZY TEXT  
> > "World"
> > S: * ESEARCH (TAG "D02") PARTIAL (1:10  
> > 19,80,85:86,95,102:103,111,113:114)
> > S: D02 OK Sort completed.
> 
> That's somewhat deceptive, since the results are in UID order, so it  
> could be misread.

Right, I didn't notice that because it was a real SORT reply that just
happened to return the results in UID order..

> What that should provide is the 10 most relevant matches, of course -  
> in this example it might be viewed that, as these are also the ten  
> first matching messages, there's some implicit "order by UID" going  
> on. A random.shuffle makes it more clear:
> 
> C: D02 SORT RETURN (PARTIAL 1:10) (RELEVANCY) UTF-8 FUZZY TEXT "World"
> S: * ESEARCH (TAG "D02") PARTIAL (1:10  
> 102,95,114,103,113,111,86,10,80,85)
> S: D02 OK Sort completed.

Changed.