When I worked on a tsearch2 application for several cr_types I noticed that there was significant change in how things used to work. I also noticed that there can be the need for lots of feature tweaking, like dictionary choice, etc. which, if changed, should force reindexing of the data, or it won't match the new queries which are prepared with the new features. Also, you can add weighting to different inputs: title could be a higher value than the body, then your query could just select certain fields to search as well.
Also missing was a boolean logic translator for the input query to the tsearch2 logic. Has any of this been integrated into the current system? Besides selecting what to search, it is important to know what the index parameters are (and if different the search will probably fail).
Just wondering what the state of search is since tseach2 was introduced.
