Andres Freund <andres@anarazel.de> writes:
> On 2018-02-27 13:36:59 -0500, Tom Lane wrote:
>> Does it seem practical to adjust TestUpgradeXversion.pm to cope with
>> this change? An alternative answer is to put back C-language stubs
>> in regress.c for the removed functions, but that seems a bit grotty.
> Could we just drop the relevant functions in the course of the test?
The change I was thinking of was to have TestUpgradeXversion.pm do that.
If we wanted to modify the back branches themselves to help with this,
what I'd be inclined to do is back-patch the regression test changes,
or some subset thereof, so that the back branches have the same idea
of which functions are in regress.so as HEAD does.
I had not intended to back-patch, since those changes were just cosmetic,
but it might be the best way to preserve the XversionUpgrade tests.
regards, tom lane