BUG #13480: RE: #11732 and also possibly #12538 - Possible serializable isolation lepessimistic locking problem

The following bug has been logged on the website:

Bug reference: 13480
Logged by: João Antunes
Email address: joao(dot)antunes(at)tagus(dot)ist(dot)utl(dot)pt
PostgreSQL version: 9.3.5
Operating system: Mac OS X
Description:

Hi Guys!

I think I might have hit problems with bug #11732, which can also be related
with #12538 - although a communication problem might have happened there.

Sorry for writing like a new email, but it's hard to contribute to a given
thread if you don't have a bug tracking system, IMO this kinda hinders this
contribution process a bit.

so,

Any news on #11732 or #12538, are those closed?

Rafael Faria: The reply to your question was basicly: please double check if
you're not doing anything wrong - did you? were you?
- In Portuguese -
Rafael Faria: A resposta que teve a sua pergunta/bug report foi basicamente:
por favor confirme que o erro não está na implementação do seu lado, havia
alguma coisa que estavam a fazer mal? conseguiu resolver o problema?

Regarding #11732 the bug is still open, right? can you confirm that in
versions < 9.1 this doesn't happen?

In my case I have a stack of JPA and ORM in the way, but if they are doing
what they should, I think I hit #11732 - for a quicker check on this, I
might go back to a version before 9.1 in case you guys tell me that bug
#11732 is non repeatable with a downgrade of postgres version, just to
actually check if this is a problem (i.e. it is the quickest way, if it
doesn't happen anymore, then probably it's that).

Unfortunately for pragmatism sake I am now considering using optimistic
locking for this (although it does bring probably unnecessary complexity on
the implementation side)

So, there Is there any 'SELECT * FROM pg_stat_activity' that also discloses
the connection's transaction isolation level? according to this:
http://stackoverflow.com/questions/28175364/how-can-i-query-the-transaction-isolation-level-of-an-existing-postgres-s
I would have to go through the logs, after enabling a more verbose way, but
this might be error prone becase AFAIK I would have to be looking for the
ommission of setting the transaction's isolation level.