ZNG dicussion

Following Ray Denenberg's e-mail that publicly announced ZNG on July 12, a
discussion was started on the ZIG list. After a week or so, this discussion
was over, partly because of the holiday season. I think however that
important things have been said, and below I try to group them in a few
categories. This summary could be useful for discussion at the ZIG in
England.
I have tried to summarize the contributions in a correct way and hopefully I
don't let anybody say things that he or she does not want to be put in the
mouth.

There is one thing that I want to bring up although I did not find any
discussion about it - maybe because the subject was touched by Denenberg
himself: The relation between ZNG and ZIG. I think we should talk about it
in Boston Spa.


Pieter van Lierop
Geac


--------------------------------
Discussion on the ZNG Initiative
--------------------------------


Presentation
------------
Some people did not like the way in which ZNG was proposed.
Zeeman talks of a "fait accompli by a really quite secretive group of
implementors".
Taylor replied that the group was created at the last ZIG meeting. (But I
was not there.)


The name
--------
Kent thinks the name "ZNG" (Z39.50 - Next Generation) is both bad and good.
Bad in that it implies it's a replacement for Z39.50 (which he do not think
it is), and good in that the name will appeal to programmers.
Dovey said the name ZNG was a mistake.


Goals
-----
What is the rationale for ZNG? (Bull)
There is no discussion of requirements. (Zeeman)
Trying to enlarge Z39.50 to a mainstream protocol is maybe not such a good
idea (Van Lierop).
Andresen asked for the long-range objective of the proposal.


ZNG and its alternatives
------------------------
There is a range of alternatives to ZNG like XML Query, XQL, XML-QL, Quilt,
why would one choose ZNG? (Bull)
Zeeman: ZNG will be unable to compete with XQL.
You can also use OAI or SQL (Sanderson).
ZNG looks very much like OAI (Futrelle).
Zeeman prefers SOAP.
There was a discussion on using XER to encode protocol messages (Futrelle,
Sanderson)


Sessions
--------
Sanderson thinks that ZNG underestimates the disadvantages of not having a
session.


Search/Present
--------------
Zeeman is against putting Search and Present together.


Dropping scan?
--------------
Van Lierop is against dropping Scan.
Kent too thinks that scan should stay.


Query language
--------------
Zeeman is against leaving RPN.
CCL is English based and especially difficult to use in a non-ASCII
character set (Bull).
There was some confusion about bib-1. Sanders was happy to find out that ZNG
drops Attribute Set bib-1. But LeVan answering Zeeman said: "Who said
anything about bib-1 going away?"


XML as Record Syntax
--------------------
Why no multiple record syntaxes are allowed? (Sanderson)
Against hard coding XML. (Van Lierop)
How does the client know what is inside the XML block? (Sanders)


ZNG and Z39.50
--------------
ZNG throws the baby out with the bath water (Davidson).
The parts omitted from ZNG form the heart of Z39.50; without them it is of
no use. (Sanderson)
Kent does not see ZNG as a replacement of Z39.50.
Ralph LeVan introduced a new argument. "Z39.50 Classic", as he calls it, is
too complicated, while "ZNG" is very simple. This provoked many reactions.
Davidson replied that you also could make a simple description of Z39.50
making it look easy. Sanderson: Changing the protocol for the sake of
programmer's ease is ludicrous. Taylor also did not agree with LeVan's
vision on the complexity of Z39.50: The protocol is not the problem, but
toolkits and libraries.

When Ray Denenberg replied briefly on July 13, before leaving on holiday,
one of things he said was: "This initiative is not intended to replace
Z39.50. It focuses on getting information to the user. The premise is that
Z39.50's strength is in business-to-business applications, and this
initiative is not aimed there."

Received on Wednesday, 26 September 2001 03:51:30 UTC