I'm not modifying any benchmark or framework. At best I will replace Mako 0.7.2 with Mako 0.7.3 in the benchmark suite since no one is historically recording the mako_v2 benchmark yet and it should be running with the newest version until we set it in stone.

On Sat, Nov 3, 2012 at 10:48 AM, Maciej Fijalkowski <fijall@gmail.com> wrote:
On Fri, Nov 2, 2012 at 8:42 PM, Brett Cannon <brett@python.org> wrote:
> Issue filed for the performance issue: http://bugs.python.org/issue16390
> With that change and running on tip of Mako on my laptop now reports 1.25x
> slower which is much better than it was. This performance issue might also
> explain why all of the regex compilation benchmarks are worse under Python
> 3.3 by a decent margin.
> On Fri, Nov 2, 2012 at 2:16 PM, Philip Jenvey <pjenvey@underboss.org> wrote:
>> lru_cache on re._compile_typed

I would like to warn you about modifying benchmarks like this (or
frameworks). Why is it relevant anyway?