
Workflow for automc:

- bugs have the string "NEEDSMC" placed in their text

- these crons are set up:

50 0,1,2 * * *  /home/jm/ftp/spamassassin/automc/RUNME.before
0 14 * * *      /home/jm/ftp/spamassassin/automc/RUNME.after

- scrape-bugzilla is run at 0850 UTC
  - and creates a new rules file "rules/70_bugzilla.cf"
  - checks it in
  - saves a temporary data file, containing:
    - the names of rules created
    - the bug numbers
    - other state

- nightly mass-check run happens
  (http://wiki.apache.org/spamassassin/NightlyMassCheck)

- output full analyzed results to somewhere like
  http://www.pathname.com/~corpus/HTML.age
  datestamped so they can be viewed (months) later

- post-comments is run at 2200 UTC
  - load the data file from scraper
  - add comment "DONEMC [commentnum]" to the bugs
    - greps out those rules' freqs and adds them to comment
    - adds link to CGI viewer app

- you must do 1 commit as the "automc" userid interactively,
  so that the credentials are cached for that user


