Re: [Exist-open] ft:optimize

[Wolfgang M. <wol...@ex...>]

> 3) It used to be that my text search was sub-second ($searchDocs[ft:query(.,
> $searchString)]) -- where $searchDocs is a sequence of searchable elements
> -- about 100K each on average.

Performance should be much better if your XPath expression provides
some context for ft:query, e.g.:

//title[ft:query(. , $searchString)]

should be many times faster than

$searchDocs[ft:query(., $searchString)]

This is because the optimizer looks at the context of the filter
expression and will take a shortcut if possible, so //title will not
be evaluated completely, while $searchDocs needs to be resolved
earlier.

I hope this information helps. If not, feel free to contact me again.
I'd like to find out why some users are having those problems. From my
experience, the Lucene index can be really fast.

Wolfgang