Re: The Emacs master is much slower than the emacs-27 branch.

[Eli Zaretskii]

> Date: Sat, 5 Dec 2020 15:56:22 +0100
> From: Ergus <spacibba@aol.com>
> Cc: "Basil L. Contovounesios" <contovob@tcd.ie>, ghe@sdf.org, acm@muc.de,
>       stephen.berman@gmx.net, emacs-devel@gnu.org
> 
> The best approach is maybe what Gregory did. Bisect the history to find
> the source of the slowdown.

Sure, if someone can afford that, it would be nice to know.  Assuming
a single commit causes it, that is (it could be several commits
instead, each one slowing down Emacs by a few percents).

When I said "not too bad", it didn't mean we should disregard it, just
that it's much less than Alan originally reported, which would be
alarming.

> Actually in my work we have a log where we register the changes and the
> commits hashes that impacted performance more than 5-7% either to
> improve or degrade it.

That requires to have a representative test suite which to use for
such benchmarking.  For Emacs this is not easy; for example, no one
said that Alan's benchmark is representative of what Emacs does most
of the time in "normal" usage.

Personally, I don't think we should even try to come up with single
benchmark.  We should instead have several, and make sure we
understand and/or fix slowdown in any one of them.  Volunteers are
welcome to work on a suite of such benchmarks, we could then put them
under test/ somewhere.