fix(3525): SET should only affect statements after it

[big-andy-coates]

Description

Fixes #3525

This commit fixes a regression that sees an old issue reappearing where by a SET statement affects not just statements that follow it, but also statements before it. The primary cause of this is the fact that ConfiguredStatement contains a reference to a mutable map containing the property overrides. This map is mutated by subsequent SET statements. The fix is to make for ConfiguredStatement to take a immutable defensive copy, as is good programming practice. This involves adding explicit handling of SET and UNSET statements, which can no long mutate the overrides in ConfiguredStatement.

Testing done

usual

Reviewer checklist

• Ensure docs are updated if necessary. (eg. if a user visible feature is being added or changed).
• Ensure relevant issues are linked (description should include text like "Fixes #")

[vcrfxia]

Thanks @big-andy-coates -- LGTM. Do you know when this regression was introduced? (Does this fix need to be back-ported?)

[big-andy-coates]

Thanks @big-andy-coates -- LGTM. Do you know when this regression was introduced? (Does this fix need to be back-ported?)

Not sure, but I don't think we do back ports do we?

[vcrfxia]

Not sure, but I don't think we do back ports do we?

If we know this regression is present on in 5.3 as well we can back-port the fix to 5.3.x so that it goes out with the next 5.3 release (in this case, 5.3.2). And similarly for earlier branches. Not sure how severe the bug needs to be in order to warrant back-porting the fix but this particular one sounds pretty bad to me.

[vcrfxia]

FWIW, tried it on 5.3.x right now and the bug is indeed present there.

Update: It's also present on 5.2.x, 5.1.x, and 5.0.x (didn't check on any earlier branches). Since it's always been present and the back-port looks like it'd be a lot of work, probably not worth the effort for this one.