Given that the SystemPhysical.xmi is closely reflects the index file
format, wouldn't removing this cause us to supply same logic in code to
to extract data from index files and present in System model format? I
imagine this needs to change again when the DDL based metadata is the
metadata model.
Would it make sense if we just materialize or cache the system tables to
get the performance you are looking for? You have mentioned this idea
before.
Thanks
Ramesh..
On Fri, 2009-10-09 at 10:34 -0400, Steven Hawkins wrote:
I am not suggesting removing System.vdb, just the SystemPhysical
model. For both Designer and Teiid the System model tables would still be visible and
usable. While removing System.vdb would be possible with Teiid (and just using code to
initialize the table/proc definitions), without significant changes to the forked metadata
code in Designer it would not be possible for them.
----- Original Message -----
From: "Ramesh Reddy" <rareddy(a)redhat.com>
To: "Steven Hawkins" <shawkins(a)redhat.com>
Cc: "teiid-users" <teiid-users(a)lists.jboss.org>,
teiid-designer-dev(a)lists.jboss.org
Sent: Friday, October 9, 2009 9:24:11 AM GMT -06:00 US/Canada Central
Subject: Re: [teiid-users] removing system physical
Designer also uses the System.vdb, are there implications there? I am
not sure what is Designer's usage pattern. Seem though, Designer does
not need to depend on it. Somebody please explain, how it is being used
in the Designer?
On Thu, 2009-10-08 at 17:27 -0400, Steven Hawkins wrote:
> Hello all,
>
> I would like to float the idea of removing SystemPhysical from the 6.3 release and
converting System to a physical model. System as a virtual model needlessly introduces
lookup tables and joins between the SystemPhysical tables. Also SystemPhysical is filled
with a lot of modeling and indexing concepts that would be best not to expose. Since
SystemPhysical is marked as non-visible, not documented for use (from within
transformations), and has all of its relevant information exposed through System anyway
the impact and potential workarounds for this removal are minimal. Let me know if there
are any concerns.
>
> Steve
> _______________________________________________
> teiid-users mailing list
> teiid-users(a)lists.jboss.org
>
https://lists.jboss.org/mailman/listinfo/teiid-users