Re: [HACKERS] proposal: session server side variables - Mailing list pgsql-hackers

From Fabien COELHO
Subject Re: [HACKERS] proposal: session server side variables
Date
Msg-id alpine.DEB.2.20.1701041423240.22281@lancre
Whole thread Raw
In response to Re: [HACKERS] proposal: session server side variables  (Fabien COELHO <coelho@cri.ensmp.fr>)
Responses Re: [HACKERS] proposal: session server side variables  (Pavel Stehule <pavel.stehule@gmail.com>)
List pgsql-hackers
> An alternative is to implement sub (nested) transactions, like Oracle 
> and MS SQL Server... but that would be quite some work.

As a complement, a search showed that IBM DB2, cited as a reference by 
Pavel, has AUTONOMOUS transactions, which looks pretty much the same thing 
as nested transactions. The documentation presents an interesting use 
security-related use case:

https://www.ibm.com/developerworks/data/library/techarticle/dm-0907autonomoustransactions/

The idea is that an application must record an attempt to access a data 
even if the attempt fails and is rolled-back.

This feature used carefully within an appropriate pattern would allow to 
ensure that if the setup transaction fails then the session status is 
FALSE. One possible inconsistency which may arise with sub xacts is that 
the status may stay FALSE while the setup has succeeded, however this on 
the safe side wrt to the security use case.

-- 
Fabien.



pgsql-hackers by date:

Previous
From: Mithun Cy
Date:
Subject: Re: [HACKERS] Cache Hash Index meta page.
Next
From: Craig Ringer
Date:
Subject: Re: [HACKERS] pg_recvlogical --endpos