Jack Christensen

Mostly looks good, but I'm getting a test failure on my machine. ``` --- FAIL: TestTimestamptzTranscode (0.02s) testutil.go:159: Param TextFormat Result TextFormat 10: expected {0001-01-01 00:00:00 +0000 UTC 2 none},...

Hmm... That simple protocol path is tricky... But at the very least it is surprising for `Value()` to return a `string` when `time.Time` is a valid `driver.Value` (and it could...

> Do you mean a special case like JSONB here? Yes. I'm still getting a strange test failure. It looks like decoding the text format is not handling time zones...

Sorry to keep nitpicking, but this latest change will permanently alter the test database when run, and this also means it requires superuser permissions. I'd prefer to avoid that. I...

As I mentioned on my comment to #1950: > One thing I would be concerned about is how to handle multiple tracers that each implement different hooks. I'm sure it...

The grouping on line 17 is a bit confusing. The mutex guards the initialization of `onCancelWasCalled` on line 42. But the assignment on line 50 and the read on line...

This is definitely an improvement and I think it would be reasonable to include something like this in puddle. But it occurred to me that with a few changes it...

What I was picturing was more like: ```go type AcquireTracer interface { AcquireStart(ctx context.Context, data AcquireStartData) context.Context AcquireEnd(ctx context.Context, data AcquireEndData) } // Maybe also have a ReleaseTracer or maybe...

I think we would also want something to trace `Release`.

> When I call JSONB.Set(src) where src is a nil map or slice, I expect it to be treated as null. It is. The problem is "which null"? That is,...