On 2/14/15, Petr Ferdus <[email protected]> wrote: > Could be fossil FTS used to search for particular text/substring hidden > anywhere > in fossil repository? >
Yes. The infrastructure is designed with that in mind, but the implementation has not been done yet. I wanted to collect more experience with the current search before getting into universal-content search. One problem to consider is that a typical full-text index requires about 20% of the space of the original document. The Fossil self-hosting repository currently holds 2.3GB of content. To build a full-text index on it all would increase the repository size from 53MB to about 500MB. About 90% of the repository would be devoted to the full-text index. That does not seem desirable. My plan to work around that is to only index the differences between successive check-ins. In other words, instead of indexing the complete text of every document, only index the changes. That will probably reduce the size of the index to be proportional to the size of the repository. And, it means when you do a search, you are only going to get hits for the particular versions where those words actually change - move into or out of the document or are in close proximity to other words that do. Of course, you cannot simply index *only* the changes. Each edit needs some context. How much context to include in the index is an open question. -- D. Richard Hipp [email protected] _______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
