Re: BUG #17385: "RESET transaction_isolation" inside serializable transaction causes Assert at the transaction end - Mailing list pgsql-bugs

Michael Paquier <michael@paquier.xyz> writes:
> On Mon, Sep 26, 2022 at 12:16:24AM -0400, Tom Lane wrote:
>> Yeah, but why?  I see no good reason why those fields need to be first.

> My reasoning on these ones is that we are most likely going to add
> more description flags in the future than new unit types.  Perhaps I
> am wrong.

Sure, but we could easily leave unused bits there.  Aligning the
units subfields on byte boundaries might result in slightly better
machine code, anyway.

            regards, tom lane