Re: Regression tests fail once XID counter exceeds 2 billion

I wrote:
> Simon Riggs <simon@2ndQuadrant.com> writes:
>> We need a function called transactionid_current() so a normal user can write

>> select virtualtransaction
>> from pg_locks
>> where transactionid = transactionid_current()

>> and have it "just work".

> That would solve that one specific use-case.  The reason I suggested
> txid_from_xid is that it could also be used to compare XIDs seen in
> tuples to members of a txid_snapshot, which is not possible now.

BTW, a pgsql-general question just now made me realize that
txid_from_xid() could have another use-case too.  Right now, there are
no inequality comparisons on XIDs, which is necessary because XIDs in
themselves don't have a total order.  However, you could
ORDER BY txid_from_xid(xmin)

and it would work, ie, give you rows in their XID order.  This could be
useful for finding the latest-modified rows in a table, modulo the fact
that it would be ordering by transaction start time not commit time.
        regards, tom lane