I used notmuch_address.py to scrape email addresses from notmuch and generate elisp to showve them into bbdb. Tab completion via bbdb is much faster than via notmuch_addresses.
Run the script, execute the results in emacs. I didn't care enough to do any merging, so records that clash with existing records are ignored. -------------- next part -------------- A non-text attachment was scrubbed... Name: scrape.py Type: text/x-python Size: 1195 bytes Desc: scrape.py URL: <http://notmuchmail.org/pipermail/notmuch/attachments/20100325/0b6b9582/attachment.py>