[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v2 1/1] xen/pci: Install vpci handlers on x86 and fix exit path



On Wed, 20 Oct 2021, Ian Jackson wrote:
> Bertrand Marquis writes ("Re: [PATCH v2 1/1] xen/pci: Install vpci handlers 
> on x86 and fix exit path"):
> > > On 20 Oct 2021, at 08:16, Jan Beulich <jbeulich@xxxxxxxx> wrote:
> > > I'm inclined to suggest s/exit/error/ in the title though (and maybe
> > > also s/path/paths/), which would be easy enough to do while committing.
> > 
> > @Ian: Please tell me if this is ok to be fixed during commit.
> 
> It's academic now, since you need a respin anyway, but I would in any
> case have preferred a new version.  I find that a much better
> workflow.  Expecting to make changes on commit pushes work onto
> committers from submitters, which is less scaleable.
> 
> And it also ends up with committer making changes without any further
> human review of any kind.  The committer must do so during an activity
> which is otherwise often shallow and administrative.  This is not the
> best way to get good results.
> 
> I believe I have made these points before, but perhaps not so clearly
> and explicitly.  They are IMO very general.  (FTAOD I'm not saying
> that making changes on commit is never appropriate, but it ought to be
> exceptional.)

Everything you wrote makes sense and your position is perfectly valid.
At the same time let me write a different perspective on the issue. All
the committers have slightly different workflows so it is natural that
things that are better for one person might not be best for another.

When I commit a patch series often I don't simply take the patches (with
all the necessary acks) and commit them. I typically do a number of
other things:
1) check that the series has all the requested changes compared to the
   previous version
2) check that nothing else was changed unexpectedly
3) run a local build test in a variety of configurations
4) run a full gitlab-ci set of tests with the patches applied

All of the above take time. So if only a small change is required, it is
much faster for me to make the change myself on commit rather than
having to go through steps 1-4 one more time again. Of course this is
only good for small changes with no impact, e.g. code style or in-code
comments, I wouldn't want to do this for anything meaningful.



 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.