----- Original Message ----- > From: "Stephen Smalley" <sds@xxxxxxxxxxxxx> > To: "Robert Kuska" <rkuska@xxxxxxxxxx>, selinux@xxxxxxxxxxxxx > Sent: Monday, July 13, 2015 4:28:42 PM > Subject: Re: [PATCH] sepolgen: add support for python3 > > On 07/13/2015 09:55 AM, Robert Kuska wrote: > > Hello everyone, > > > > this is my first mail to this mailing list so I would like to > > introduce myself a bit. > > > > I am Robert Kuska, member of python-sig in Fedora and co-owner > > of change Python3 as default. In Fedora many packages depends > > on policycoreutils which ships python modules from selinux, the > > problem is (at least for us as we want to have default installation > > clean of Python2) that not all of them work with Python3. Therefore > > I've decided to make a patch(es) for all modules (except for GUI) > > to make them work with python3. > > > > As a first I've picked a sepolgen as it seemed it will be the hardest. > > Please note that I have no previous experience with coding for selinux > > project. > > > > You can find my patch as an attachment to this mail, as it is a rather > > big one (I didn't split the patch because it fixes mostly same syntax > > errors). > > > > I have tested the patch with python2.7 and python3.4. I had > > no issue with python2.7 except for test_module.TestModuleCompiler.test > > which was failing even before my patch. > > Do you have selinux-policy-devel installed? My bad, after installing the devel package all tests pass (on python2). Thank you. > > > > > Following failures (number vary, from 1 to 6) were seen with python3.4: > > > > https://paste.fedoraproject.org/243817/36795382/ > > > > The origin of these failures lies within change in hash function of sets > > and dicts, hash now uses random seed in python3 and tests rely on > > frozen order of items which is imho bad. I didn't fix those tests > > as I wanted to know firstly if this is just a issue with tests and such > > frozen order is not expected. > > I didn't write sepolgen, but it is certainly the case that the order in > which permissions or types are listed in a policy rule makes no > difference to its semantics. Jason Zaman said the same thing on irc, I just wanted to be sure, I can send another patch with fixed tests. > > > -- Robert Kuska {rkuska} _______________________________________________ Selinux mailing list Selinux@xxxxxxxxxxxxx To unsubscribe, send email to Selinux-leave@xxxxxxxxxxxxx. To get help, send an email containing "help" to Selinux-request@xxxxxxxxxxxxx.
