RFR: Section on backporting [v7]

Thu Dec 2 23:05:27 UTC 2021

On Thu, 2 Dec 2021 22:27:31 GMT, Jesper Wilhelmsson <jwilhelm at openjdk.org> wrote:

> If it's desired to redo the backport filing a REDO bug might be sufficient, but if it's found to be too complex to fix whatever issues caused the backout a REDO bug won't be filed, leaving us in a situation where the main issue looks like it has been backported to 11u when in fact the change is not available in 11u.

That's a good point: Without a redo bug it will be confusing to anyone looking at it. It's also might be a little confusing to remove the record, but perhaps not as confusing as leaving it there.

> Even if the REDO bug is filed, keeping the backported-by-link to the backed out backport would cause confusion as the main issue still looks like it's already backported. And if the second backport is pushed to the same update release there will be two backport issues with the same fix version.

The redo will use a different main bug IDs, so I don't see that this adds confusion. Your earlier point is probably enough of a reason to do what you suggest.

Btw, I wouldn't suggest deleting the backport relationship for a bug that has been backed out in mainline (shortly) after it was backported, and thus backed out in the update release as well. What you propose makes sense only for bugs that remain in mainline but not in one or more of the update releases. This should be very rare for the reasons you mentioned earlier, but it can happen.

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

PR: https://git.openjdk.java.net/guide/pull/66