Peter, with issue 18805 fixed, what do you think about removing the provisional status from the ipaddress API?

Aside from issue 19157 (where I'm not sure the behaviour even needs changing), the other open ipaddress issues appear to be docs related or RFEs.

At this point, making ipaddress subject to the usual backwards compatibility guarantees seems fairly low risk.

Works for me.

I was going to say that there was still the issue with the mask parsing which turned out to be more difficult than simply applying the ipaddress-py patch but it looks like you've managed to get that applied. I spent a few hours banging my head on the keyboard before.

Yeah, I quickly gave up on trying to *apply* that netmask parsing patch for issue 18805, and instead just reimplemented it using the original as a guide (and, most importantly, as the source of the new test cases). The slightly different testing setup in the stdlib version also meant I could ensure that IPInterface was fixed in addition to IPNetwork.

Issue 19157 was the only one that gave me pause, but we won't change that until 3.5 anyway, and it is well within the limits of the kind of changes we consider acceptable in feature releases.

I'll make the change to the 3.4 docs (including What's New) to remove the disclaimer.

New changeset 2b9c10f3b969 by Nick Coghlan in branch 'default':
Close #20563: Declare ipaddress API stable
http://hg.python.org/cpython/rev/2b9c10f3b969