This one was a nice example: https://gerrit.cloudera.org/#/c/4186/

I think a good benchmark is that small patches like that should only take
1-2 review iterations once a contributor has the code style and process
down.

On Fri, Dec 9, 2016 at 11:34 AM, Jim Apple <[email protected]> wrote:

> +1. Thanks for coming up with this idea and sending out this note,
> Tim. I think this is the right move.
>
> On Fri, Dec 9, 2016 at 11:14 AM, Tim Armstrong <[email protected]>
> wrote:
> > Hi All,
> >   I just wanted to follow up on the previous emails about starting the
> code
> > review process for your Impala patches. We'd really like the process to
> go
> > as smoothly as possible
> >
> > One thing we really need to figure out is how we might test and maintain
> > PPC code. We only really have access to x86 build infra and it isn't
> > feasible for us to maintain PPC code that we can't test. It's a burden on
> > contributors who want to touch the code and gain confidence they're not
> > breaking anything, and also on PPC devs who will likely get accidentally
> > broken repeatedly.
> >
> > I'm also concerned making sure that the code review process goes as
> > smoothly as possible, since each iteration can take a lot of time,
> > especially if we're working across different time zones. We have a fairly
> > high bar for code quality and maintainability, and in our experience it
> > takes a bit of time for contributors to calibrate for that and post code
> > reviews that can get through the process smoothly.
> >
> > I'd suggest that you should identify some small platform-agnostic fixes
> in
> > your patchsets and start off with posting code reviews for those. That
> way
> > we can make progress and get you ramped up on the code review process
> > before we've resolved the more difficult PPC testing questions.
> >
> > https://cwiki.apache.org/confluence/display/IMPALA/
> Contributing+to+Impala
> > is our main reference for the process.
> >
> > Thanks,
> > Tim
>

Reply via email to