[PATCH 1 of 4] revlog: use absolute_import

[Matt Mackall]

On Wed, 2015-07-15 at 14:33 -0700, Gregory Szorc wrote:
> On Wed, Jul 15, 2015 at 8:52 AM, Augie Fackler <raf at durin42.com> wrote:
> 
> > On Tue, Jul 14, 2015 at 01:51:10PM -0700, Gregory Szorc wrote:
> > > # HG changeset patch
> > > # User Gregory Szorc <gregory.szorc at gmail.com>
> > > # Date 1436894831 25200
> > > #      Tue Jul 14 10:27:11 2015 -0700
> > > # Node ID 5154b8421ebd670d6155659c5811f193f5fe1555
> > > # Parent  35fa7c77c754aa4d156c42abfdb61ca178468872
> > > revlog: use absolute_import
> >
> > This patch looks like straightforward code cleanup to me, and I think
> > we should just apply it now.
> >
> > The rest of the series seems reasonable, but I think those make sense
> > for early in the next cycle rather than at the end of this one.
> >
> 
> I understand the reasoning for wanting to hold the file handle changes.
> However, this line of work all started around trying to fix issue 4750 and
> there is some anecdotal evidence that the problem is more widespread at
> Mozilla than initially believed. I'd very much like to get *something* in
> 3.5 to address the issue. Unfortunately, a full and proper fix is
> definitely not possible for 3.5 if this series is out of scope...

I've taken patch 1, but (as discussed on IRC), I think we should go back
to our earlier callback+flush approach to fix the issue that triggered
this in 3.5, then revisit the more general issue here post-3.5.

-- 
Mathematics is the supreme nostalgia of our time.