[Python-ideas] Exploiting type-homogeneity in list.sort() (again!)

[Elliot Gorokhovsky]

On Sun, Mar 5, 2017 at 11:31 PM Tim Peters <tim.peters at gmail.com> wrote:

>
> The best approach is indeed to pass the function pointer to every
> location that needs it.  Note that a MergeState struct is already
> created per sort invocation, That isn't a file global for much the
> same reason.
>

Right. It's a real shame, because the patch as it stands right now is
extremely surgical, but there's clearly no way around it. There are some
functions that don't take in MergeState (e.g. gallop_left) but that call
ISLT/IFLT, so I think I'll just add a compare function pointer parameter to
all the function calls. I mean, the diff will be a lot hairier, which might
make the patch more difficult to review, but when it comes down to it the
code complexity won't really increase, so overall I don't think this is the
end of the world.

Thanks so much for pointing this out!

P.S. Is it OK if I close my current issue on the bug tracker and open a new
issue, where I'll post the revised patch? The writing on my current issue
uses the old, less-rigorous benchmarks, and I feel it would be less
confusing if I just made a new issue and posted the correct
benchmarks/discussion at the top. The current issue doesn't have many
comments, so not much would be lost by just linking to it from the new
issue. If this violates bug tracker etiquette, however, :) I'll just post
the revised patch on my current issue.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-ideas/attachments/20170306/76765ae6/attachment.html>