cvs commit: src/sys/conf files src/sys/netinet tcp_ofld.c tcp_ofld.h tcp_var.h toedev.h src/sys/sys socket.h

[Tom Rhodes]

On Fri, 14 Dec 2007 02:11:51 +0000 (GMT)
Robert Watson <rwatson at FreeBSD.org> wrote:

> 
> On Thu, 13 Dec 2007, Kip Macy wrote:
> 
> > On Dec 13, 2007 5:30 PM, Robert Watson <rwatson at freebsd.org> wrote Sadly, 
> > often the only way to get a real discussion going is to make the immediacy 
> > of it relevant. To date I haven't made any material structural changes to 
> > TCP, I've only added the hooks that will be needed. As requested by another 
> > I will add some commentary on the purpose of each of the individual hooks to 
> > the header file.
> 
> I'd certainly agree with the observation that it takes immediacy to force 
> review and discussion to take place.  However, I think it's also the case that 
> continuous review of a significant WIP is very time-consuming for the 
> reviewers.  By structuring the review process a bit (i.e., identifying 
> specific spots in the design, implementation, etc, where seeking review makes 
> sense and there's a fairly fixed work product for someone to look at rather 
> than a rapidly-moving target in which any comments are rapidly invalidated), I 
> find I tend to receive much more productive reviews from others.  Certainly, 
> "The attached patch is going into the tree on/about date X" is the most 
> effective technique, other than just committing the change, to prompt 
> review...

The patch size (line count) matters a lot as well.  It's much easier
to get a 10 line review than a 100 line review.  Especially if, as
you noted, it's a significant WIP.

-- 
Tom Rhodes