I just tried John's patch, and: - it still applies without problem (except for a bit of offset) - I can confirm that it actually adds test coverage for leapdays() function (bringing calendar coverage from 71% to 72%). I think it would be good to apply it to py3k branch. Cheers, Sandro
Applied in r87590. I threw in an extra test for a multi-leapyear-range. Since there was no reason not to, I backported it to 3.1 in r87591 and 2.7 in r87592. In the latter two commits I also backported the issue 9342 patch. Thanks for the patch, John, and for the review, Sandro.