Re: Nested Transactions, Abort All

Alvaro Herrera wrote:
> On Sat, Jul 10, 2004 at 09:46:56AM +1200, Oliver Jowett wrote:
>
>>Alvaro Herrera wrote:
>>
>>
>>>We can't actually release anything (commit the subtransactions), because
>>>they may be savepoints established after that point, and they are
>>>logically "inside" the previously established ones. At RELEASE we can't
>>>really release -- we just lose the name and thus the opportunity to
>>>rollback to it.
>>
>>The 2003 draft claims that RELEASE SAVEPOINT invalidates savepoints
>>subsequent to the RELEASE:

(that should read "subsequent to the released savepoint")

> In our case, invalidating a savepoint does not mean we can release its
> resources. We can only do that if it's the latest defined savepoint.

I don't understand why this is true if the invalidation comes from a
RELEASE statement. I understand the problems with savepoint name reuse
invalidating an earlier savepoint -- we do have to keep the earlier txn
open in that case.

Say I have:

SAVEPOINT s1
-- work 1
SAVEPOINT s2
-- work 2
RELEASE SAVEPOINT s1 -- Invalidates s1 and s2

Can't we translate that to:

begin subtransaction s1
-- work 1
begin subtransaction s2
-- work 2
commit subtransaction s1 -- and implicitly s2

We don't need to keep subtransaction s2 open -- we will never need to
roll it back as the RELEASE of s1 invalidates it.

What am I missing?

-O