[Xapian-tickets] [Xapian] #304: Calculate LHS weight earlier in AND_MAYBE to avoid calling RHS in some cases

[Xapian]

#304: Calculate LHS weight earlier in AND_MAYBE to avoid calling RHS in some
cases
-------------------------+--------------------------------------------------
 Reporter:  richard      |        Owner:  olly    
     Type:  enhancement  |       Status:  new     
 Priority:  normal       |    Milestone:          
Component:  Matcher      |      Version:  SVN HEAD
 Severity:  normal       |   Resolution:          
 Keywords:               |    Blockedby:          
 Platform:  All          |     Blocking:          
-------------------------+--------------------------------------------------

Comment(by richard):

 My thinking was that this optimisation would be helpful in situations
 where tight bounds on the weight of the RHS for parts of the posting list,
 are known.  For example, if we kept wdf bounds for each chunk of the
 posting source, it would be possible to avoid having to check the
 existence of a posting if the weight was below the bound for the relevant
 chunk of the posting source.

 Also, if the RHS was a phrase operator, it is possible to determine the
 weight of the phrase before checking if the phrase actually existed - this
 could avoid needing to access the positional information in some cases.

 The new method would actually be a check2() method, rather than a
 skip_to2() method, I think - if the document-specific minweight supplied
 to it was above the known maxweight for the chunk of the posting list, the
 check2() method would simply return false, to indicate that the document
 is known to not be weighted high enough.

-- 
Ticket URL: <http://trac.xapian.org/ticket/304#comment:2>
Xapian <http://xapian.org/>
Xapian