[11u] Sconpe of Review ... was RFR(S): 8241234: Unify monitor enter/exit runtime entries.

[Andrew Haley]

On 09/09/2020 16:49, Lindenmaier, Goetz wrote:
> It is pointless to ask reviewers to judge the risk
> because reviews are only done if the change had to be adapted.

That's an excellent point. In such a case, approvals are the only
time risk is considered. However, if a patch applies cleanly I
suppose it's probably less risky. Although in some cases that in't
true.

> There are complex changes that just apply clean and thus are
> downported without review.
>
> Judging the risk is clearly a thing of the downporter. This is
> formulated in Rule 1 of Oracles's Updates description :
> http://openjdk.java.net/projects/jdk-updates/approval.html , and
> also mentioned in step 6 of jdk11us description:
> https://wiki.openjdk.java.net/display/JDKUpdates/How+to+contribute+a+fix
> This is supposed to help the maintainer to decide about the risk.

Yes, of course the person doing the porting must consider the risk.
But so should the person reviewing it.

> The major task of the review is to make sure the downported change
> is correct, i.e. has the same effect on 11 as the original one.#

I disagree. Just like a review of a change to head, the reviewer must
consider whether the justification put forward by the submitter is
sufficient.

> Nevertheless, if there is a  reviewer and he feels bad about a
> change, he should communicate his concerns!

Yes, that should happen.

It's everybody's duty, all the time, to consider risk. We owe our
users no less than that.

-- 
Andrew Haley  (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671