Created on 2010-08-22 22:24 by brett.cannon, last changed 2011-12-02 23:33 by brett.cannon. This issue is now closed.
| Messages (3) | |||
|---|---|---|---|
| msg114714 - (view) | Author: Brett Cannon (brett.cannon) *  |
Date: 2010-08-22 22:24 | |
Importlib does not use any OS-level protections to gain exclusivity when opening a file like import.c does through open_exclusive. It probably should, though, when writing bytecode else one might end up with corrupt code. That's bad as bad marshal data is a flat-out import failure and not simply glossed over. Plus if I don't do this now I will just end up getting a bug report that test_multiprocessing is randomly failing on the buildbots because of this issue since that "precious" little test seems to love to ferret out concurrency issues in importlib. |
|||
| msg114716 - (view) | Author: Jesse Noller (jnoller) * |
Date: 2010-08-22 23:02 | |
Nice. |
|||
| msg148780 - (view) | Author: Brett Cannon (brett.cannon) * |
Date: 2011-12-02 23:33 | |
Fixed by Issue13146. |
|||
| History | |||
|---|---|---|---|
| Date | User | Action | Args |
| 2011-12-02 23:33:41 | brett.cannon | set | status: open -> closed resolution: out of date messages: + msg148780 superseder: Writing a pyc file is not atomic stage: test needed -> |
| 2010-08-22 23:02:54 | jnoller | set | nosy:
+ jnoller messages: + msg114716 |
| 2010-08-22 22:24:11 | brett.cannon | create | |