home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.cs.arizona.edu
/
ftp.cs.arizona.edu.tar
/
ftp.cs.arizona.edu
/
tsql
/
doc
/
tsql.mail
/
000125_rts _Thu May 13 21:16:28 1993.msg
< prev
next >
Wrap
Internet Message Format
|
1996-01-31
|
3KB
Received: from boojum.CS.Arizona.EDU by optima.cs.arizona.edu (5.65c/15) via SMTP
id AA06647; Thu, 13 May 1993 21:16:29 MST
Date: Thu, 13 May 1993 21:16:28 MST
From: "Rick Snodgrass" <rts>
Message-Id: <199305140416.AA11476@boojum.cs.arizona.edu>
Received: by boojum.cs.arizona.edu; Thu, 13 May 1993 21:16:28 MST
To: tsql@cs.arizona.edu
Subject: Shashi's comments
Shashi makes some good points about both the taxonomy and the instance.
I wish he had made these points when these were being actively designed
and discussed.
As it is, it is now only four weeks from the workshop. If we start
redesigning the instance and the taxonomy now, there is no way we
will have a draft benchmark ready for the workshop.
Recall that this effort started several months ago. There has been
a great deal of constructive technical discussion about all the
portions of the benchmark, from the objectives, to the schema, to
the instance, to the taxonomy. The traffic on tsql related solely
to the benchmark has averaged approximately one message per day. Many
people have made general and specific suggestions that have been
incorporated into the present document.
Christian has tried to keep us on schedule by moving on to the next
topic once a general consensus has been reached. He recommended some
time ago that we consider the instance and the taxonomy finalized,
and move on to the last task, that of writing the queries. There was
little disagreement on this at the time.
I suggest that we stick with this schedule. After the workshop, we'll
have a better understanding both of the first version of the benchmark,
because it will actually exist, as well as other aspects of temporal
DB infrastructure. At that point, we can revisit all portions of
the benchmark. Others also have concerns that should be addressed at
that time. The second version of the benchmark will be much stronger
as a result.
An alternative is to simply decide that the schedule of completing an
initial version of the benchmark is simply unrealistic. I am strongly
opposed to that alternative, because I believe that without a schedule,
we would still be discussing task 1, and a benchmark would never emerge.
However, I also strongly feel that this effort needs to be by consensus,
so I put the following question on the floor, for (hopefully quick)
discussion and resolution:
Should we continue to generate an initial draft of the benchmark
by June 16, or should we consider that goal unrealistic, and
resume discussion of the objectives, schema, instance, and
taxonomy before moving on to the queries at some later date?
I request that all (current and potential) authors of the benchmark
respond quickly with their views on this question.
Thank you.