launchpad-dev team mailing list archive

[Martin Pool <mbp@xxxxxxxxxxxxx>]

2010/1/11 Graham Binns <graham@xxxxxxxxxxxxx>:
> Writing this as I'm dashing out the door (so I'll respond to the rest
> of your mail later). The algorithm and the reasoning behind it is now
> on the BugHeat page (I meant to put it there on Friday and didn't;
> sorry).

Thanks, that helps.

istm that if you want hotness to be an overall measure of how much
things matter, you might want importance to be included too, but not
in such a way that it overrides others.  I guess experience will tell.
 Perhaps that would be a good replacement for specially handling
bug-control or team members, since only special teams can set
importance.  In your table you could try

critical +1000
high +500
wishlist -300

istm that if Launchpad counted subscribed/affected users across
duplicates correctly (for which there is a bug) then you might not
need to care much about duplicates.  In the trivial common case of
there being N duplicates each reported by one user who is marked
subscribed/affected, the math would come out just the same.

many bugs have lots of pointless bulk subscribers.  presumably these
don't count?

It looks like the current list includes recency of updates - is that
just because it is the secondary sort key?

Perhaps the scaling factors would be clearer with some worked examples:

known critical, no dupes or affected users - higher
recently reported, not yet triaged, 10 affected users - pretty high
...

I realize a lot of discussion has already gone.

-- 
Martin <http://launchpad.net/~mbp/>