[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Antoine Pitrou added the comment: Python 3.3 now has per-module import locks, and PyImport_ImportModuleNoBlock simply redirects to PyImport_ImportModule. See issue9260. -- resolution: -> duplicate stage: -> committed/rejected status: open -> closed superseder: -> A finer grained imp

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Changes by Martin Dengler : -- nosy: +mdengler ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.pytho

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Changes by Eric Snow : -- nosy: +eric.snow ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.python.or

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Changes by Jesús Cea Avión : -- nosy: +jcea ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.python.o

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Changes by Raymond Penners : -- nosy: +pennersr ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.pyth

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Changes by MunSic JEONG : -- nosy: +ruseel ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.python.or

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Antoine Pitrou added the comment: > Pretty much anything mentioned in PEP 302 is what is being protected. Alright, my question was more along the lines of "what *needs* to be protected". The import lock is currently like the GIL: a very simple (and therefore suboptimal) answer to a complex prob

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Nick Coghlan added the comment: Added Guido as well, to see how much he remembers from back when Christian's patch went in. We may want to take this over to python-dev... -- nosy: +gvanrossum ___ Python tracker _

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Nick Coghlan added the comment: Pretty much anything mentioned in PEP 302 is what is being protected. The one which causes most of the grief is the fact that we need to stick a partially initialised module into sys.modules (so the module can find itself if it needs to), then take it out again

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Antoine Pitrou added the comment: Is there some documentation somewhere of which global variables the import lock is meant to protect? Perhaps it wouldn't need to be held during the whole module initialization process. -- nosy: +pitrou ___ Python t

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Nick Coghlan added the comment: OK, finally tracked down some of the history here. Issue 1567 is the one where Christian developed the relevant patch which was then committed as r59678. Issue 7980 is also potentially related (although that's apparently a Mac OS X crash involving time.strptime

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Graham Dumpleton added the comment: Nick, there is no main module in the same way there is when using the Python command line. Ie., there is no module that has __name__ being __main__. The closest thing is the WSGI script file which holds the application object for the WSGI application. There

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Nick Coghlan added the comment: I'm curious - do the mod_wsgi problems go away if you arrange for the main module to run with the import lock held? Or do they turn into deadlocks? -- ___ Python tracker ___

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Alexander Belopolsky added the comment: Composed before reading Nick's comment: """ This issue is broader in scope than just the time module example that is given, so I am not taking this over. With respect to datetime and time modules, there are several related issues I am working on which

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Nick Coghlan added the comment: Interesting. Your example code doesn't run afoul of any of the "don't do this" flags relating to import that I put into the threading docs, so it seems like a fair complaint to me. As Christian pointed out on c.l.p, the idea with this API is to turn deadlocks

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Changes by Mark Lawrence : -- nosy: +ncoghlan ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.python

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

Changes by Antoine Pitrou : -- nosy: +belopolsky, brett.cannon versions: +Python 3.2 -Python 2.6 ___ Python tracker ___ ___ Python-bugs

[issue8098] PyImport_ImportModuleNoBlock() may solve problems but causes others.

New submission from Graham Dumpleton : Back in time, the function PyImport_ImportModuleNoBlock() was introduced and used in modules such as the time module to supposedly avoid deadlocks when using threads. It may well have solved that problem, but only served to cause other problems. To illus