Message205811
The real issue here is that the test used to determine whether -S was passed or not is outdated: instead of checking sys.flags.no_site, it checks whether 'site' is in sys.modules. This is no longer a valid test, since site's side effects are contained within site.main, which is only run if no_site is False. In fact, distutils imports symbols from site unconditionally, as do a couple of test modules; hence why test_site currently fails running `python -S Lib/test`: test_distutils (and others) run before test_site and cause site to be present in sys.modules.
I think we should move away from a toplevel SkipTest, though; some of the tests may be applicable whether -S is passed or not, though of course the ImportSideEffectTests would not be. |
|
Date |
User |
Action |
Args |
2013-12-10 13:36:55 | zach.ware | set | recipients:
+ zach.ware, Arfrever, r.david.murray, vajrasky |
2013-12-10 13:36:55 | zach.ware | set | messageid: <1386682615.82.0.362254555475.issue19828@psf.upfronthosting.co.za> |
2013-12-10 13:36:55 | zach.ware | link | issue19828 messages |
2013-12-10 13:36:55 | zach.ware | create | |
|