mxnet-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Madan Jampani <madan.jamp...@gmail.com>
Subject Re: Formalize Committer Proposal and Application Procedure
Date Fri, 04 Aug 2017 15:14:53 GMT
There is a middle ground here. Instead of saying someone either has full
committer privileges or zero, an alternative is to have scope of ownership
start small and localized to modules or source folders where their primary
contributions currently lie. For example, there are folks who contributed
full languages bindings, or very good examples/tutorials.

Over time, depending on the scope and complexity of their contributions
they can be nominated to have their commit privileges broadened or even
become core committers. Core committers have full commit privileges.

Irrespective of whether some one is committer or not, we should all be
using the PR process and opening up contributions for review/feedback.

Madan



On Fri, Aug 4, 2017 at 5:04 AM, Isabel Drost-Fromm <isabel@apache.org>
wrote:

> On Fri, Aug 04, 2017 at 12:27:16PM +0100, Chiyuan Zhang wrote:
> > Suppose we lower the standard or completely remove the formal standard
> for
> > committers, then we could probably be able to get more committers from
> the
> > first type. But that might not necessarily be good to us
>
> Can you elaborate your reasoning here? (I'm not implying that I agree or
> disagree with you, I just want to understand where this fear is coming
> from.)
>
>
> > having people that could either contribute relatively important
> components
> > or provide longer term commitment to the project. But on the other hand,
> > having a standard for committers do not (I hope) discourage the first
> type
> > of contributors to contribute PRs.
>
> Let me tell you a little campfire story: Back in the old days of Mahout we
> implicitly had a relatively high bar for becoming a committer. People
> thought
> that in order to become committer they would have to contribute substantial
> patches, often full new algorithm implementations.
>
> What the project really needed were a lot of work polishing, optimising,
> cleaning, making easier to use, documenting etc.
>
> Due to the perception of requiring substantial contributions to get the
> reward of becoming committer however we never received much of the latter.
>
>
> Lesson learnt for me: The way you setup your reward systems greatly
> influences which kind of help your project will receive.
>
>
> Isabel
>
> --
> Sorry for any typos: Mail was typed in vim, written in mutt, via ssh (most
> likely involving some kind of mobile connection only.)
>

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