From: thhal at mailblocks.com (Thomas Hallgren)
Date: Wed, 25 May 2005 11:15:36 +0200
Subject: [Pljava-dev] Re: the relation ID of the table on which the
	trigger occured.
In-Reply-To: <4294383B.3080507@lewczuk.com>
References: <d71a0u$lbt$1@sea.gmane.org>	<42942E42.9090005@mailblocks.com>
	<thhal-0xMZvA9X0zTYdkH7uFjvrCJvJ6lNMHk@mailblocks.com>
	<thhal-0xcZvA9f0zTYJXwu2y5ljtHShqmiCii@mailblocks.com>
	<4294383B.3080507@lewczuk.com> <42944238.6060609@mailblocks.com>
	<thhal-0utpvA+/0zTYNy1B8A6mMewJeBLuiFz@mailblocks.com>
Message-ID: <thhal-0utpvA/H0zTYRP8VB2BsoxS0/nuwSqQ@mailblocks.com>

Marek Lewczuk wrote:

> Thomas Hallgren napisa?(a):
>
>> The TriggerData.getTableName() will give you the name of the relation. 
>
> True, but it returns only table's name, without schema name - table's
> name can't be used as a true identifier of given table (because table
> e.g. "pljavaexample" can exists in many schemas) and because of that
> relation ID is needed.

That's a bug in that case. The table must of course be qualified.

>> The TriggerData interface is designed to be as database agnostic as 
>> possible (to allow other vendors to use the same interface) and the 
>> relation OID is very PostgreSQL specific.
>
> I understand, however that can lead to many other losts on the pljava's
> functionality - PostgreSQL is very complex and it contains many unique
> features, which are not available in other RDBMS.

No, there will be no functionality loss. Just a clear separation of 
concern. The idea is that if you want to write portable code, then stick 
with the public interfaces. If you want to do PostgreSQL specific stuff, 
use pljava.internal stuff.

Regards,
Thomas Hallgren