Comment 4 for bug 537735

Revision history for this message
Jean-Baptiste Lallement (jibel) wrote :

I think that's the same kind problem than bug 554180. Thats because you're querying 2 different database with nearly the same data. xapian return both records e.g
$ delve -d /var/lib/apt-xapian-index/index -d ./data/xapian/ -t 'brasero'
Posting List for term `brasero' (termfreq 5, collfreq 30): 2990 [Brasero] 14487 [brasero-common] 38457 [brasero] 41695 [python-brasero] 43553 [rhythmbox-plugin-cdrecorder]

The 1rst record (Brasero) is from ./data/xapian/ (s-c cache) and the 3rd (brasero) is from axi.

You should use a collapse key, for instance the package name or an md5 of the description and eliminate duplicate documents at search time based on that key. It would be even simpler and better to use only one index and do that at index time.

I've the feeling that this redundant db will cause some trouble in the future.