Re: Subject: CFV: New JDK 10 Committer: Erik Österlund

Jim Graham james.graham at oracle.com
Thu Jun 22 19:44:04 UTC 2017


Vote: abstain

I am abstaining because I just wanted to comment on the process here, but I don't know anything about Erik's contributions.

On 6/22/17 10:02 AM, mark.reinhold at oracle.com wrote:
> The key question to answer when someone is nominated to a Committer role
> is: Do you trust them to use their push access in a responsible fashion?

This is the key distinction here.

The only real consideration here is "are they aware of the process and do they follow it".

Since all contributions must be peer reviewed before they go in, all we really need is for them to understand how to get 
those reviews and to understand and honor that they need them before they can do the push.  Honesty and familiarity with 
the process is the most key here.  We could have someone who has no deep understanding of the code, but all they really 
want to do is fix docs and as long as they've gone through the review process 8 or so times to fix their typos, they are 
pretty much good to go on doing their own pushes at that point - once they get their required reviews on each subsequent 
fix.

The expectation is not that they'll fix 10 typos with reviews, then we grant them Committer status and they can push on 
their own and suddenly they rewrite the world and push it without any reviews.  This vote just says "do we think this 
person can be trusted to follow the process (and are they familiar with the process)?".

Reviewer status is where we want them to have a deep understanding so that they don't look at a fundamental change to 
the locking states and inter-module communication paths and say "Approved" because they didn't spot any typos and the 
patch compiles without warnings.

Committer should be relaxed to "went through the review process N times and understands it" and Reviewer should involve 
"significance"...

> Speaking for myself, I've seen enough of Erik's work that I think he can
> be trusted, so I'm going to vote "yes".

Trust in following the process is really the only important thing here, no?  It isn't so much the content of the work, 
as it is their understanding of the review process.

Possibly "trust in understanding their own limitations" might be important too.  But, since they are still subject to 
reviews, even that isn't really necessary as long as they'll defer to reviewers who tell them that they've exceeded 
their limitations...

			...jim


More information about the jdk10-dev mailing list