One thing that still needs fixing, as mentioned on IRC: in the test, the two distroseries have the same id. That happens to produce the right results here, but it does upset the bookkeeping a little bit.
The ordering of the ids in the test comparison seems a little arbitrary (5, 7, 6, 4 IIRC). If possible, use ordering-insensitive comparisons where ordering is not important and test ordering separately. Less brittle, easier to comprehend.
Thanks for making these changes. Much better now!
One thing that still needs fixing, as mentioned on IRC: in the test, the two distroseries have the same id. That happens to produce the right results here, but it does upset the bookkeeping a little bit.
The ordering of the ids in the test comparison seems a little arbitrary (5, 7, 6, 4 IIRC). If possible, use ordering- insensitive comparisons where ordering is not important and test ordering separately. Less brittle, easier to comprehend.
But now I'm nit-picking. Well done!
Jeroen