This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author ashkop
Recipients ashkop, ezio.melotti, flox, serhiy.storchaka
Date 2015-03-02.15:24:04
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <>
The issue occurs only if C implementation of _warnings is available. It is caused by re-use of global filters variable from _warnings when warnings is being re-imported.
So warnings modifies _warnings.filters on first import. Then, when you import warnings again it uses already modified version of _warnings.filters and adds same filter for the second time.

I attach a simple patch that prevents duplicates in filters. I'm not sure if this is a best way to fix the problem, but it works. It certainly affects performance of filterwarnings(), but there shouldn't be a lot of filters anyway.

Making filters a set() will solve the problem also, but will require API changes. Although I didn't find anywhere in documentation that warnings.filters should be a list, someone might rely on its .append() method.
Date User Action Args
2015-03-02 15:24:04ashkopsetrecipients: + ashkop, ezio.melotti, flox, serhiy.storchaka
2015-03-02 15:24:04ashkopsetmessageid: <>
2015-03-02 15:24:04ashkoplinkissue18383 messages
2015-03-02 15:24:04ashkopcreate