[PATCH] pure: write a really lazy version of pure indexObject. On PyPy this version performs

[Matt Mackall]

On Tue, 2016-05-10 at 10:34 +0200, Maciej Fijalkowski wrote:
> On Wed, May 4, 2016 at 6:15 PM, Martin von Zweigbergk
> <martinvonz at google.com> wrote:
> > 
> > On Sun, Apr 24, 2016 at 4:32 AM, Maciej Fijalkowski <fijall at gmail.com>
> > wrote:
> > > 
> > > # HG changeset patch
> > > # User Maciej Fijalkowski <fijall at gmail.com>
> > > # Date 1461496898 -10800
> > > #      Sun Apr 24 14:21:38 2016 +0300
> > > # Branch stable
> > > # Node ID de5eed1fbf88ed5a514eecdeeba83ccb26ec5a1e
> > > # Parent  2d3837a4bded5362f26f91033c0a83376c207593
> > > pure: write a really lazy version of pure indexObject. On PyPy this
> > > version performs
> > > reasonably well compared to C version. There is potential for improvements
> > > by storing
> > > extra as a condensed struct too.
> > Can you be more specific than "performs reasonably well"? Can I ask
> > you to find some command that exercises this code and include timings
> > for the pure version before, pure version after, and C version. Does
> > that make sense?
> The problem here is that there is no command that would have "pure"
> indexObject-bound performance, so the gains are minimal.

Commands: no. Features: yes. Basically any DAG-oriented revset can hit this
pretty heavily. There's a revset benchmark suite in contrib.

-- 
Mathematics is the supreme nostalgia of our time.