Re: gsoc, oprrest function for text search

Heikki Linnakangas wrote:
> Jan Urbański wrote:
>> Here's a WIP patch implementing an oprrest function for tsvector @@
>> tsquery and tsquery @@ tsvector.
>>
>> The idea is (quoting a comment)
>> /*
>>  *  Traverse the tsquery preorder, calculating selectivity as:
>>  *
>>  *   selec(left_oper) * selec(right_oper) in AND nodes,
>>  *
>>  *   selec(left_oper) + selec(right_oper) -
>>  *      selec(left_oper) * selec(right_oper) in OR nodes,
>>  *
>>  *   1 - select(oper) in NOT nodes
>>  *
>>  *   freq[val] in VAL nodes, if the value is in MCELEM
>>  *   min(freq[MCELEM]) / 2 in VAL nodes, if it is not
>
> Seems reasonable.
>
>>  *
>>  * Implementation-wise, we sort the MCELEM array to use binary
>>  * search on it.
>>  */
>
> Would it be possible to store the array in sorted order, to avoid
> sorting it on every invocation of tssel?

It's being stored sorted on frequencies, like so:
[('dog', 0.9), ('cat', 0.8), ('sheep', 0.7)]
and I need it sorted on elements for bsearch().

I don't know if it's OK to break the rule that statistical data is
stored sorted on freqneucies. If so, then ts_typanalyze() would have to
change and do one more qsort() before storing the result.

--
Jan Urbanski
GPG key ID: E583D7D2

ouden estin