At UDS-O in Budapest a few weeks ago we had a session (video) on DEX: an initiative to improve collaboration between Debian and its derivatives. There's an Ubuntu DEX subproject, and this is what the session was about, thinking up ideas.
I thought of something that could possibly yield some results --- my action from the session was:
[laney] measure relative uploads between ubuntu and debian and see if anything interesting pops up: TODO
It seems to me that if we could tell which packages were uploaded with the most frequency to Ubuntu with respect to Debian, then we may be able to figure out why this is and if there's any way we can do the work in Debian instead.
Clearly for this to happen we are going to need data! And where better to house a load of data than the Ultimate Debian Database itself?
So I wrote some scripts to download the Ubuntu changes, handily made available as mbox archives, chop them around a bit and eventually split them out in a format that UDD can understand. Lucas Nussbaum then kindly integrated the output into UDD itself, and the end result is that an as-near-as-possible complete history of Ubuntu uploads is now available for your creative querying. For example, Wednesday sees the most uploads and Sunday the fewest:
udd=> SELECT EXTRACT (DOW FROM date) AS dow, COUNT(EXTRACT (DOW FROM date)) AS count FROM ubuntu_upload_history GROUP BY dow ORDER BY count DESC; dow | count -----+------- 3 | 26739 2 | 26699 1 | 25361 5 | 23497 4 | 22154 6 | 9687 0 | 8752 (7 rows)
I'll lovingly craft a script to generate the data we need for
the DEX project in the next few days. But I hope that the data
proves useful for other things too. The three tables you want to
look at (with
\d+ at a psql prompt) are
ubuntu_upload_history_launchpad_closes (LP bugs closed
by uploads) and
bugs closed, probably not so interesting).
Information on how to use UDD yourself is on this wiki page.