Created on 2017-07-10 19:11 by brett.cannon, last changed 2017-07-11 01:22 by ncoghlan.
| Messages (2) | |||
|---|---|---|---|
| msg298077 - (view) | Author: Brett Cannon (brett.cannon) *  |
Date: 2017-07-10 19:11 | |
Based on all the work currently going into fixing the locking situation for imports, it's probably a good idea to expose it publicly so others can use it. |
|||
| msg298098 - (view) | Author: Nick Coghlan (ncoghlan) * |
Date: 2017-07-11 01:22 | |
Based on r126574323">https://github.com/python/cpython/pull/2646/files#r126574323, I'm actually thinking along different lines: I'm starting to think we should move the current sys.modules to sys._modules, and make sys.modules a synchronised wrapper that does the following for all get/set/delete operations: # Acquires the global import lock # Acquires the relevant module lock # Releases the global import lock # Mutates sys._modules # Releases the module lock That said, I'm not sure that would be sufficient to give the desired level of atomicity, so exposing ModuleLockManager may actually be a better approach. |
|||
| History | |||
|---|---|---|---|
| Date | User | Action | Args |
| 2017-07-11 01:22:17 | ncoghlan | set | messages: + msg298098 |
| 2017-07-10 19:11:35 | brett.cannon | set | nosy:
+ ncoghlan, eric.snow |
| 2017-07-10 19:11:26 | brett.cannon | set | dependencies: + importlib: _find_and_load() race condition on sys.modules[name] check |
| 2017-07-10 19:11:21 | brett.cannon | create | |