Message5813

Author	adam.spiers
Recipients	adam.spiers, bpedman, matt_brinkley, pjenvey, rkurin
Date	2010-06-16.11:59:30
SpamBayes Score	0.0027578843
Marked as misclassified	No
Message-id	<1276689572.19.0.702297178326.issue1522@psf.upfronthosting.co.za>
In-reply-to

Content
I think I was wrong about the repeated compilation. According to my debugger, makeProxy() generates a new org.python.proxies.foo class every time the code is run, not every time it is compiled. I don't know why this is necessary, but I think this should mean these proxies "belong" to a unique PythonInterpreter and that it's safe to remove these proxies from PyType.class_to_type during finalization of that PythonInterpreter. I'll test that now. I would like to echo Matt's request for an explanation of how this is all supposed to work, because this part of the Jython code is extremely complex. Is there a Jython internals document available anywhere?

I think I was wrong about the repeated compilation.  According to my debugger, makeProxy() generates a new org.python.proxies.foo class every time the code is run, not every time it is compiled.  I don't know why this is necessary, but I think this should mean these proxies "belong" to a unique PythonInterpreter and that it's safe to remove these proxies from PyType.class_to_type during finalization of that PythonInterpreter.  I'll test that now.

I would like to echo Matt's request for an explanation of how this is all supposed to work, because this part of the Jython code is extremely complex.  Is there a Jython internals document available anywhere?

History
Date	User	Action	Args
2010-06-16 11:59:32	adam.spiers	set	messageid: <1276689572.19.0.702297178326.issue1522@psf.upfronthosting.co.za>
2010-06-16 11:59:32	adam.spiers	set	recipients: + adam.spiers, pjenvey, matt_brinkley, bpedman, rkurin
2010-06-16 11:59:31	adam.spiers	link	issue1522 messages
2010-06-16 11:59:30	adam.spiers	create