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 rhettinger
Recipients jdemeyer, rhettinger, tim.peters
Date 2018-09-21.01:21:21
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1537492882.45.0.956365154283.issue34751@psf.upfronthosting.co.za>
In-reply-to
Content
ISTM, you're being somewhat aggressive and condescending.  I'll hand you off to Tim who is far better at wrestling over the details :-)

From my point of view, you've made a bold and unsubstantiated claim that Python's existing tuple hash has collision issues with real code.  Only a contrived and trivial integer example was shown.

On my end, I made an honest effort to evaluate the suggestion.  I read your post, 
looked up Bernstein hash, revisited the existing C code and found that everything it is doing makes sense to me.  There was a comment indicating that empirical tests were run to get the best possible constants.  Also, I disassembled the code to verify that it is somewhat efficiently compiled.

Your responded that I'm being "silly" and then you reversed my decision to close the issue (because there is no evidence that the hash is broken).  Accordingly, I'm going to hand it over to Tim who I'm sure will know the history of the existing code, can compare and contrast the various options, evaluate them in the context of typical Python use cases, and assess whether some change is warranted.
History
Date User Action Args
2018-09-21 01:21:22rhettingersetrecipients: + rhettinger, tim.peters, jdemeyer
2018-09-21 01:21:22rhettingersetmessageid: <1537492882.45.0.956365154283.issue34751@psf.upfronthosting.co.za>
2018-09-21 01:21:22rhettingerlinkissue34751 messages
2018-09-21 01:21:21rhettingercreate