|
From: SourceForge.net <no...@so...> - 2003-06-12 09:34:45
|
Bugs item #753154, was opened at 2003-06-12 11:34 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=551954&aid=753154&group_id=78018 Category: com Group: None Status: Open Resolution: None Priority: 5 Submitted By: Andrea Babini (ababini) Assigned to: Nobody/Anonymous (nobody) Summary: memory leak wrapping object having _typelib_guid_ attribute Initial Comment: Problem experimented using win32all build 152, Python 2.2.2. The attached .zip file contains: A) TestServer.tlb that defines the IPippo interface B) CPippo.py that implements IPippo referring the typelib using _com_interfaces and _typelib_guid_ attributes C) PyPippo.py that implements IPippo using _public_method_ attribute D) CTest.py that reproduce the problem using CPippo E) PyTest.py that is the same as CTest.py but use PyPippo and doesn't have the leak F) universal.py as modified by me to reduce the size of the leak. My changes are only a workaround to reduce the leak in my specific use (only one interface for object, no lcid and typelib version support). Moreover I think that with this change the performances are improved as well. The steps to reproduce the problem are: 1. COM registrations 1.1 python regtlb.py TestServer.tlb 1.2 python CPippo.py 1.3 python CTest.py 2. Test to highlight the leak (it use CPippo class) 2.1 python CTest.py ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=551954&aid=753154&group_id=78018 |
|
From: SourceForge.net <no...@so...> - 2004-04-22 07:50:44
|
Bugs item #753154, was opened at 2003-06-12 19:34 Message generated for change (Comment added) made by mhammond You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=551954&aid=753154&group_id=78018 Category: com Group: None >Status: Closed >Resolution: Fixed Priority: 5 Submitted By: Andrea Babini (ababini) Assigned to: Nobody/Anonymous (nobody) Summary: memory leak wrapping object having _typelib_guid_ attribute Initial Comment: Problem experimented using win32all build 152, Python 2.2.2. The attached .zip file contains: A) TestServer.tlb that defines the IPippo interface B) CPippo.py that implements IPippo referring the typelib using _com_interfaces and _typelib_guid_ attributes C) PyPippo.py that implements IPippo using _public_method_ attribute D) CTest.py that reproduce the problem using CPippo E) PyTest.py that is the same as CTest.py but use PyPippo and doesn't have the leak F) universal.py as modified by me to reduce the size of the leak. My changes are only a workaround to reduce the leak in my specific use (only one interface for object, no lcid and typelib version support). Moreover I think that with this change the performances are improved as well. The steps to reproduce the problem are: 1. COM registrations 1.1 python regtlb.py TestServer.tlb 1.2 python CPippo.py 1.3 python CTest.py 2. Test to highlight the leak (it use CPippo class) 2.1 python CTest.py ---------------------------------------------------------------------- >Comment By: Mark Hammond (mhammond) Date: 2004-04-22 17:49 Message: Logged In: YES user_id=14198 Thanks! I found the underlying memory leaks, and fixed them. Your sample no longer leaks at all, even though I didn't implement the cache in your patch. Note that by using gencache.EnsureModule(), you get the same benefits of the cache (and not even take the hit each time the process starts 0 just once as a .py is generated) ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=551954&aid=753154&group_id=78018 |
|
From: SourceForge.net <no...@so...> - 2004-04-22 08:24:03
|
Bugs item #753154, was opened at 2003-06-12 19:34 Message generated for change (Comment added) made by mhammond You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=551954&aid=753154&group_id=78018 Category: com Group: None Status: Closed Resolution: Fixed Priority: 5 Submitted By: Andrea Babini (ababini) Assigned to: Nobody/Anonymous (nobody) Summary: memory leak wrapping object having _typelib_guid_ attribute Initial Comment: Problem experimented using win32all build 152, Python 2.2.2. The attached .zip file contains: A) TestServer.tlb that defines the IPippo interface B) CPippo.py that implements IPippo referring the typelib using _com_interfaces and _typelib_guid_ attributes C) PyPippo.py that implements IPippo using _public_method_ attribute D) CTest.py that reproduce the problem using CPippo E) PyTest.py that is the same as CTest.py but use PyPippo and doesn't have the leak F) universal.py as modified by me to reduce the size of the leak. My changes are only a workaround to reduce the leak in my specific use (only one interface for object, no lcid and typelib version support). Moreover I think that with this change the performances are improved as well. The steps to reproduce the problem are: 1. COM registrations 1.1 python regtlb.py TestServer.tlb 1.2 python CPippo.py 1.3 python CTest.py 2. Test to highlight the leak (it use CPippo class) 2.1 python CTest.py ---------------------------------------------------------------------- >Comment By: Mark Hammond (mhammond) Date: 2004-04-22 18:24 Message: Logged In: YES user_id=14198 I meant to mention - I added "pippo" to the standard test suite, including the IDL etc. It currently doesn't do much beyond a simple leak test, but is a great framework for further tests - thanks! ---------------------------------------------------------------------- Comment By: Mark Hammond (mhammond) Date: 2004-04-22 17:49 Message: Logged In: YES user_id=14198 Thanks! I found the underlying memory leaks, and fixed them. Your sample no longer leaks at all, even though I didn't implement the cache in your patch. Note that by using gencache.EnsureModule(), you get the same benefits of the cache (and not even take the hit each time the process starts 0 just once as a .py is generated) ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=551954&aid=753154&group_id=78018 |
