Re: transactional shared variable in postgresql

Why dont you try to make 'username' part of your query.
While auditing updates and inserts, the username can be some how made
a part of the tuple being inserted. Just a thought!

--Imad
www.EnterpriseDB.com

On 11/18/06, Rajesh Kumar Mallah <mallah(dot)rajesh(at)gmail(dot)com> wrote:
> Hi ,
>
> Is there any way to set a variable from a web application (using dbi/perl ,
> libpq etc),
> and access the same variable from a C trigger inside a transaction ?
>
> the %_SHARED hash available in plperl provides only session level isolation
> and
> does not suit the requirement.
>
>
> Original problem:
>
> we want to setup table auditing using table_log or audittrail projects.
> the triggers used in these systems uses the current database user
> to log to the audit tables..
>
> In our webapps, we use the same username to connect to the database
> for all kind of updates. Hence we are not able to makeout whoo modified
> what . However at application level we have different "userid" for different
> users of the system. we want to somehow pass this "userid" to the databasee
> server and accesss it from the triggers that implement the audit functions.
>
> we thought of using a table of single row and single column to store
> the userid but we are concerned over the performance due to locking
> issues.
>
> Regds
> Mallah.
>