Doron Cohen/Haifa/[EMAIL PROTECTED] wrote on 14/01/2007 23:04:27: > David <[EMAIL PROTECTED]> wrote on 14/01/2007 20:08:05: > > > thanks, How do Lucene give each document an ID when the document is > added? > > Is the document ID unchanged until the document is deleted? > > > > Not exactly. > > When the first doc is added, it is assigned id 0. > Next one assigned id 1, etc. > When a doc is deleted, it is first only marked as such. > So if there are 10 docs they have ids 0 to 9. > > Now doc 2 and 4 are deleted, - there is no change in ids. > Next doc added is assigned id 10. > > Now if/when the segment containing the deleted docs is merged, all info on > those docs is really removed, and docids are modified to remove any holes > in the numbering - result is: 0 docs with ids 0 to 8. Now, next doc added
That shuld be 9 docs (not 0...) > gets id 9. > > Btw, segments are merged either as result of explicit call to optimize(), > or implicitly following addDoc or indexWriter.close() (and depending on > Lucene's merge policy). > > Docids are therefore internal, with unstable values. > > See also the FAQ - http://wiki.apache.org/jakarta-lucene/LuceneFAQ > Especially "When is it possible for document IDs to change?" > Also take a look at the http://lucene.apache.org/java/docs/fileformats.html - at the "Document Numbers" section. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
