Michael Musgrove

Sorry @jmfinelli I didn't get around to this today - I will have another go at scheduling it tomorrow.

I am still okay to merge this PR.

/cc @stuartwdouglas Do you recall your thinking process when choosing REQUIRES_NEW for the default Transaction Semantic?

I've also see the advice to also set `max_connections` to the same value as `max_prepared_transactions`.

> Is this PR still necessary? I don't think so but we need @zhfeng to decide.

> SHA-224 hash is indeed 28 bytes long, but the UTF-8 String created from it is quite bigger, Interesting but it should easy to fix, perhaps by using a different...

> > I don't understand this statement. > > Look how it gets initialized in TxControl without further proof! My point was that I didn't understand what you wrote because...

I don't recall you being this uncooperative and obstinate in our previous interactions so let's avoid conversation and I will just use your bug description, thanks for the initial report.

@zhfeng we could add that property but it is transactionally unsafe so we I doubt we'd add it.

Even allowing a single one-phase aware resource to join an XA transaction containing two-phase aware resources is transactionally unsafe. Beyond that, allowing two such resources is asking for trouble and...