Ahh - this is why I didn't find this regression: I was only testing the initial
unbundle code path, which doesn't exercise this changing appending code.
I'm glad this perf issue is due to something wonky in hg and not some
filesystem issue.
> On Dec 15, 2016, at 11:01, Durham Goode
> On Dec 15, 2016, at 2:01 PM, Durham Goode wrote:
>
> # HG changeset patch
> # User Durham Goode
> # Date 1481828418 28800
> # Thu Dec 15 11:00:18 2016 -0800
> # Node ID 6db5cb70cecbd50835f7c8f28770643279001e76
> # Parent