FWD: OBJS comments on the PIDL submission

Venu gave me permission to forward this message publicly.

-Rolf
------- Start of forwarded message -------
Sender: venuv@w3.org
Date: Thu, 11 Feb 1999 11:02:15 -0800
From: Venu Vasudevan <venu@objs.com>
X-Accept-Language: en
To: rnelson@w3.org
Subject: OBJS comments on the PIDL submission
Content-Type: multipart/mixed;
 boundary="------------C1DF78CD97ECF479ED5696D8"

This is a multi-part message in MIME format.
- --------------C1DF78CD97ECF479ED5696D8
Content-Type: multipart/alternative;
 boundary="------------FA296FC17088169A43815F7B"


- --------------FA296FC17088169A43815F7B
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit


Below are our comments on the recent PIDL (Personalized Information
Description Language) submission to W3C.  I can be reached at
venu@objs.com for any subsequent discussions or clarifications.

Regards,
Venu.

- -----------------------------------------------------------------------------

PIDL is an interesting application of database view definition
technology to page personalization on the web. We have the following
comments relating to the current proposal:


   *  PIDL is an interesting  blend of view virtualization and view
     materialization.
        o In that it stores "views as queries" it is a virtual view
          approach
        o  in that it allows agents to "process" (i.e materialize) a
          view, and then adds the materialized results to the document
          it is a view materialization scheme. In fact  it is worse, as
          the resulting document is (original content + materialized
          views  + view specifications).
   * PIDL does not seem to allow transmission of the materialized view
     only (minus the original contents), something I would expect to be
     best suited for thin  clients. The argument that sending fat pages
     (content+union-of-all-views) makes  multicasting more efficient, is
     true, but the scheme comes with other warts. Seems to me that
     PIDL is dictating a policy on content clustering which is not
     univresally advantageous. Instead, it could be policy neutral and
     provide mechanisms for a processing agent to either cluster the
     materialized views with the original content (or not).
   *  PIDL seems to allow each user to view every other user's view
     specification (no access control). This could be problemmatic in a
     corporate environment, e.g. Do I really want marketing to know what
     I am NOT looking at?
   *  PIDL does not make a strong case for its existence once XQL is in
     place. If it is an interim step waiting for XQL, why not inlcude
     this as a sub-activity of XQL to avoid standards fragmentation.
     While DOM may have its drawbacks, the other possibility would be to
     include sort and filter primitives in the DOM API.
   * Not sure I see the need for a comparison between PIDL and RDF.
     PIDL is a view specification, RDF is about page semantics. What's
     the connection?



- --------------FA296FC17088169A43815F7B
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit

<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
&nbsp;
<br>Below are our comments on the recent PIDL&nbsp;(Personalized Information
Description Language) submission to W3C.&nbsp; I&nbsp;can be reached at
venu@objs.com for any subsequent discussions or clarifications.
<p>Regards,
<br>Venu.
<p>-----------------------------------------------------------------------------
<br>PIDL is an interesting application of database view definition technology
to page personalization on the web. We have the following comments relating
to the current proposal:
<br>&nbsp;
<ul>
<li>
&nbsp;PIDL is an interesting&nbsp; blend of view virtualization and view
materialization.</li>

<ul>
<li>
In that it stores "views as queries" it is a virtual view approach</li>

<li>
&nbsp;in that it allows agents to "process" (i.e materialize) a view, and
then adds the materialized results to the document it is a view materialization
scheme. In fact&nbsp; it is worse, as the resulting document is (original
content + materialized views&nbsp; + view specifications).</li>
</ul>

<li>
PIDL does not seem to allow transmission of the materialized view only
(minus the original contents), something I would expect to be best suited
for thin&nbsp; clients. The argument that sending fat pages (content+union-of-all-views)
makes&nbsp; multicasting more efficient, is true, but the scheme comes
with other warts. Seems to me that PIDL&nbsp;is dictating a policy on content
clustering which is not univresally advantageous. Instead, it could be
policy neutral and provide mechanisms for a processing agent to either
cluster the materialized views with the original content (or not).</li>

<li>
&nbsp;PIDL seems to allow each user to view every other user's view specification
(no access control). This could be problemmatic in a corporate environment,
e.g. Do I really want marketing to know what I am NOT looking at?</li>

<li>
&nbsp;PIDL&nbsp;does not make a strong case for its existence once XQL
is in place. If it is an interim step waiting for XQL, why not inlcude
this as a sub-activity of XQL to avoid standards fragmentation. While DOM&nbsp;may
have its drawbacks, the other possibility would be to include sort and
filter primitives in the DOM&nbsp;API.&nbsp;</li>

<li>
Not sure I&nbsp;see the need for a comparison between PIDL&nbsp;and RDF.&nbsp;
PIDL is a view specification, RDF is about page semantics. What's the connection?</li>
</ul>

<p>&nbsp;</html>

- --------------FA296FC17088169A43815F7B--

- --------------C1DF78CD97ECF479ED5696D8
Content-Type: text/x-vcard; charset=us-ascii;
 name="venu.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Venu Vasudevan
Content-Disposition: attachment;
 filename="venu.vcf"

begin:vcard 
n:Vasudevan;Venu
tel;work:602.755.7581
x-mozilla-html:TRUE
org:OBJS Consulting
adr:;;1942 E.Todd Drive;Tempe;AZ;85283;USA
version:2.1
email;internet:venu@objs.com
x-mozilla-cpt:;4576
fn:Dr. Venu Vasudevan
end:vcard

- --------------C1DF78CD97ECF479ED5696D8--
------- End of forwarded message -------


-- 
| Rolf Nelson (rolf@w3.org), Project Manager, W3C at MIT
|   "Try to learn something about everything
|             and everything about something."  --Huxley

Received on Tuesday, 16 February 1999 09:28:38 UTC