Author xdegaye
Recipients eryksun, haypo, martin.panter, mba, serhiy.storchaka, steve.dower, xdegaye, xiang.zhang
Date 2017-05-20.14:12:02
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1495289522.87.0.243547076551.issue29619@psf.upfronthosting.co.za>
In-reply-to
Content
Good point.
PyLong_FromLong() is faster than PyLong_FromLongLong(), IMO the fastest should be used when it is possible.
PyLong_FromUnsignedLong() and PyLong_FromUnsignedLongLong() are almost identical, they implement the same algorithm. The first one uses unsigned long (on the stack and for a register) while the other uses unsigned long long, so the difference in performance may be in the order of nano seconds on current processors. Should only PyLong_FromUnsignedLongLong() be used then or should this be left as is for consistency with the handling of PyLong_FromLong() ?

Another point (not related) is that the d_ino member of the DirEntry structure is of type ino_t which is unsigned and PR 1666 should be corrected to use unsigned instead in the last modifications made by the PR in os_DirEntry_inode_impl(). I will push this change when a decision has been taken on the point raised by Martin.
History
Date User Action Args
2017-05-20 14:12:02xdegayesetrecipients: + xdegaye, haypo, martin.panter, serhiy.storchaka, eryksun, steve.dower, xiang.zhang, mba
2017-05-20 14:12:02xdegayesetmessageid: <1495289522.87.0.243547076551.issue29619@psf.upfronthosting.co.za>
2017-05-20 14:12:02xdegayelinkissue29619 messages
2017-05-20 14:12:02xdegayecreate