reef-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dongjoon Hyun <dongj...@apache.org>
Subject Re: "The overhead of reviewing and testing"
Date Wed, 27 Jan 2016 07:51:58 GMT
Thank you for your generous comments. Actually, what I meant to focus is
the vice versa.

"The reviewer (in a review cycle) should focus on only the PR code itself
like Blind Paper Review Process."

- Why does a reviewer consider a contributor's other PRs?
- Why does a reviewer ask something beyond the scope of the PR?
  (As we know, contributors are not REEF's or PMC's full-time employees.)

I dream that an ideal environment where any tiny contributions from anyone
are welcome equally(here, blindly) anytime.

Although the content of PR does not improve REEF, a reviewer should say a
warm comment like 'Sorry, but thanks for making a PR', and give -1. I think
that's natural in Open Source Communities. (The PR code will be withdrawed
or closed.)

I know that REEF community consists of strong developers and has warm
atmosphere in general. I wrote here because I was surprise about the
concept of `overhead`. I really wanted to listen others' opinions, too.
Thank you, Markus.

Dongjoon.



On Tue, Jan 26, 2016 at 10:58 PM, Markus Weimer <markus@weimo.de> wrote:

> Hi,
>
> thanks for bringing this up. It is an interesting, different angle on the
> review process. We've had this discussion from the other side before:
> contributors felt that the review cycle put a impediment onto them.
>
> I believe both arguments have a kernel of validity, but I also believe
> that reviews are super-crucial not only for code quality, but even more so
> for a shared understanding of the code. As you said, every committer and
> PMC member should use the opportunity to stay current with REEF's
> development via the review process. To be honest, that is often my reason
> for picking specific PRs: I want to know what's happening in that part of
> the code.
>
> That said, I think we should use every tool and idea available to remove
> friction from the process. Basic coding standard checks and test runs
> shouldn't be the work of human brains, but CPUs :) The work Mariia and you
> have been doing goes a long way towards that. There are some gaping holes
> (*ahem* .NET tests) which should be addressed soon.
>
> But I am rambling: What do other's think? Is the PR review process too
> burdensome? What (beyond the basic continuous integration setup) can we do
> to make it easier?
>
> Thanks again for bringing this up!
>
> Markus
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message