Michael Paquier <michael@paquier.xyz> writes:
> On Tue, Aug 28, 2018 at 05:47:47PM -0400, Tom Lane wrote:
>> I didn't want to backpatch further than v11 without a test case that would
>> work in those branches, and I lacked one. If you've got out-of-core code
>> you could verify it with, please do that and back-patch further.
> Was there any need to patch v11 with that actually?
I figured that patching v11 was safe without further testing, because
HEAD has hardly diverged from that. Previous branches probably should
get tested in some fashion before back-patching, and I didn't have a
good test case, so I didn't. But I suspect that a back-patch would
be worthwhile because (a) external modules would like to rely on such
infrastructure and/or (b) we might like to back-patch test cases for
contrib modules. However ...
> I have reviewed the modules I have, and actually it seems that I would
> not need much of that for a back-patch. One reason being that most of
> my TAP tests need pg_regress so as nodes can be initialized so this
> needs an external installation anyway.
... if there's other missing pieces then neither (a) nor (b) is very
compelling.
regards, tom lane