Re: Automatically ROLLBACK after fall in *ABORT STATE*

From: hstenger(at)adinet(dot)com(dot)uy
To: pgsql-hackers(at)postgresql(dot)org
Subject: Re: Automatically ROLLBACK after fall in *ABORT STATE*
Date: 2000-07-28 20:14:49
Message-ID: 3981E9B9.98D1F533@adinet.com.uy
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

hstenger(at)adinet(dot)com(dot)uy wrote:
> We all know PostgreSQL inability to handle exceptions. Inside a transaction, any
> error will make it fall in *ABORT STATE*, and any further action not being
> ROLLBACK, ABORT or COMMIT, will ask for a ROLLBACK, ABORT or COMMIT statement. I
> want to know what exact code is executed inside PostgreSQL, when a ROLLBACK
> statement is issued. Knowing that, I will modify backend/tcop/postgres.c, to do
> the necesary calls to ROLLBACK, inmediately after AbortCurrentTransaction().
> This will make PostgreSQL behave at least just like IBM's DB2, which rolls back
> automatically after an error.

I'm new to gdb.

I compiled PostgreSQL --enable-debug, and installed it.
Logged in as postgres superuser.
Ran postmaster -i as foreground.
Opened a second shell.
Opened ddd-gdb debugger.
Opened postmaster binary with symbols.
Opened a third shell.
Run psql template1. This makes the postmaster fork.
>From the gdb, attach to the forked postmaster, the one attending the psql
connection, and start it.

Now I run commands from psql, and expect the debugger to follow the backend
resolution of my queries. But nothing moves.

I want to see what happens in the backend when I enter ROLLBACK in psql.

Please, help me.

Regards,
Haroldo.

--
----------------------+------------------------
Haroldo Stenger | hstenger(at)ieee(dot)org
Montevideo, Uruguay. | hstenger(at)adinet(dot)com(dot)uy
----------------------+------------------------
Visit UYLUG Web Site: http://www.linux.org.uy
-----------------------------------------------

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Malcolm Beattie 2000-07-28 20:53:34 Re: Security choices...
Previous Message Henry B. Hotz 2000-07-28 18:39:27 Re: mac.c