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 belopolsky, benjamin.peterson, jyasskin, mark.dickinson, pitrou, rhettinger, serprex, stutzbach
Date 2010-09-02.00:23:43
SpamBayes Score 4.927532e-06
Marked as misclassified No
Message-id <1283387028.15.0.919042469028.issue9155@psf.upfronthosting.co.za>
In-reply-to
Content
I'm marking this one as rejected.  The timings have shown mixed results (some favorable, some not).  In general, there is a bias against expanding the number of opcodes because 1) there aren't that many codes available, 2) it grows the size of the switch-case (which has been problematic from some compilers on various systems), and because it complicates other optimization efforts such as the peephole optimizer or in other tools that depend on bytecode (such as the Unladen Swallow LLVM project, Psyco, or ByteCodeHacks). 

Also, the potential improvement is very small because it removes only one indirection out of long chain of events involved in a comparison.  Most of the time is consumed in abstract dispatch to a concrete comparison method and it that method itself.
History
Date User Action Args
2010-09-02 00:23:48rhettingersetrecipients: + rhettinger, mark.dickinson, belopolsky, pitrou, jyasskin, benjamin.peterson, stutzbach, serprex
2010-09-02 00:23:48rhettingersetmessageid: <1283387028.15.0.919042469028.issue9155@psf.upfronthosting.co.za>
2010-09-02 00:23:45rhettingerlinkissue9155 messages
2010-09-02 00:23:43rhettingercreate