Re: [GIT/PATCH v5] xen network backend driver

[Ian Campbell <Ian.Campbell@eu.citrix.com>]

On Mon, 2011-03-14 at 22:22 +0000, David Miller wrote:
> From: Ian Campbell <Ian.Campbell@eu.citrix.com>
> Date: Mon, 14 Mar 2011 22:17:40 +0000
> 
> > Would you consider a commit patch which imports the baseline driver as
> > is from xen.git (this builds and is functional but needed clean up)
> > followed by the cleanup series? I made sure the cleanup part builds and
> > works at each step as I went (but I will run through it again to be
> > sure).
> > 
> > There is real value (to me at least) in keeping the cleanup separate and
> > being able to (even manually) associate the first upstream commit with
> > an equivalent point in the historical code.
> 
> Do you have any idea what that thing is going to do for poor souls trying
> to bisect?
> 
> GIT is going to hop in and out of your line of development for any GIT
> bisect that traverses any period of time in which those driver commits
> exists.
> 
> It's too messy.

It's exactly the same as any other patch series in this respect though,
isn't it?

But... I'd forgotten that up until "xen: netback: Make dependency on
PageForeign conditional" I was running with a local non-upstreamable
branch containing the PageForeign infrastructure merged which I removed
once I removed the PageForeign dependency was gone, so the series isn't
actually fully buildable upstream like I thought.

So I'll resend a single patch version of the driver today.

> You can keep your tree online somewhere to publish the history, just
> like we do for the old bitkeeper import and similar.

Sure.

Ian.