[PATCH STABLE] obsolete: fix n^2 marker computation behavior

[Matt Mackall]

On Mon, 2016-03-07 at 16:10 +0100, Pierre-Yves David wrote:
> 
> On 02/23/2016 07:05 PM, Augie Fackler wrote:
> > On Tue, Feb 23, 2016 at 02:54:52PM +0100, Pierre-Yves David wrote:
> > > 
> > > 
> > > On 02/23/2016 02:19 PM, Pierre-Yves David wrote:
> > > > # HG changeset patch
> > > > # User Durham Goode <durham at fb.com>
> > > > # Date 1454629084 28800
> > > > #      Thu Feb 04 15:38:04 2016 -0800
> > > > # Branch stable
> > > > # Node ID 61be787ee62ea5910b0e55b5b5743079fbea1457
> > > > # Parent  1bcb4f34b9f91a2e330966182f691664fbada1bc
> > > > # Available At http://hg.netv6.net/marmoute-wip/mercurial/
> > > > #              hg pull http://hg.netv6.net/marmoute-wip/mercurial/ -r
> > > > 61be787ee62e
> > > > obsolete: fix n^2 marker computation behavior
> > > 
> > > This is a grafting on stable of a change from durham currently on default.
> > > This is hitting people in the wild and the change is small enough for
> > > stable
> > > in my opinion.
> > 
> > I concur, but I'll leave that to mpm (cc'd) to decide.
> 
> Gentle ping.

The duplicate summary caused this to be swallowed by my "have I already dealt
with this?" tools.

For future reference, I think avoidable quadratic (or worse!) behavior is fair
to consider an outright bug and appropriate for stable, especially if we see it
manifest in 100s+ overhead.

What I don't like to see on stable is performance tweaking with small constant
factors. A 2x speed-up is awesome.. but not worth risking regression on stable
for.

I've manually grafted this here, thanks.

-- 
Mathematics is the supreme nostalgia of our time.