Message12334

Author jeff.allen
Recipients fwierzbicki, jamesmudd, jeff.allen, stefan.richthofer
Date 2019-01-22.23:19:37
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1548199177.43.0.0681076386326.issue2654@roundup.psfhosted.org>
In-reply-to
Content
I have begun to write notes in a sharable form, about some of the difficult aspects of Jython, including one about this: https://github.com/jeff5/inside-jython/blob/master/source/import.rst

I think a sufficient answer here, short of the "ideal" revision of modules and import, is to keep (something like) impAttr, but ensure it only finds Java classes.

Writing regression tests for import is difficult as one has to set up temporary folders on sys.path, and change to the interpreter is permanent. The tests we have from CPython test that things that should work, do work. Less often do they test that things that shouldn't happen (like magical imports) do not happen. The same is true of the tests for #2455.
History
Date User Action Args
2019-01-22 23:19:37jeff.allensetmessageid: <1548199177.43.0.0681076386326.issue2654@roundup.psfhosted.org>
2019-01-22 23:19:37jeff.allensetrecipients: + jeff.allen, fwierzbicki, stefan.richthofer, jamesmudd
2019-01-22 23:19:37jeff.allenlinkissue2654 messages
2019-01-22 23:19:37jeff.allencreate