RFR: More on testing [v2]

[Sean Mullan]

On Tue, 21 Feb 2023 19:55:28 GMT, Jesper Wilhelmsson <jwilhelm at openjdk.org> wrote:

>> I think the 24 hour rule was hotspot specific. Other areas do very quick turn-arounds.
>
> I can't say that I have studied what happens in reality in other areas, but the 24 hour text was added as a general rule in "Working with pull requests" in PR #79 which is one of the more reviewed PRs in the Guide. Several developers from all areas of the JDK was involved and no one had any issues with the rule. In fact, the only comment about that particular guidance was from a person active in the libraries area and he suggested to add the sentence about holidays etc to say that sometimes even 24 hours isn't enough. So it seems to me that this is an accepted rule in general. Wether people actually follow the rule may be a different story.

FWIW, I think the 24 hour "rule" is a good one for all component teams, except I think it should be more like a best practice, rather than a hard rule. Exceptions would be for trivial fixes and fixes for build errors or tier 1 or tier 2 failures where a quicker turnaround is necessary.

-------------

PR: https://git.openjdk.org/guide/pull/98