home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.cs.arizona.edu
/
ftp.cs.arizona.edu.tar
/
ftp.cs.arizona.edu
/
tsql
/
doc
/
tsql.mail
/
000107_nls@cse.iitb.ernet.in _Wed May 5 06:36:14 1993.msg
< prev
next >
Wrap
Internet Message Format
|
1996-01-31
|
5KB
Received: from relay2.UU.NET by optima.CS.Arizona.EDU (5.65c/15) via SMTP
id AA29979; Wed, 5 May 1993 06:36:14 MST
Received: from spool.uu.net (via LOCALHOST.UU.NET) by relay2.UU.NET with SMTP
(5.61/UUNET-internet-primary) id AA07152; Wed, 5 May 93 09:36:15 -0400
Received: from sangam.UUCP by spool.uu.net with UUCP/RMAIL
(queueing-rmail) id 093044.25943; Wed, 5 May 1993 09:30:44 EDT
Received: by sangam.ncst.ernet.in (4.1/SMI-4.1-MHS-7.0)
id AA22498; Wed, 5 May 93 18:52:03+0530
Received: from kailash.cse.iitb.ernet.in by iitb.ernet.in
SENDMAIL Version (4.1/SMI-4.1-MHS-7.0)
id AA16700; Wed, 5 May 93 18:39:21+0530
Received: by kailash.cse.iitb.ernet.in (4.1/SMI-4.1)
id AA18047; Wed, 5 May 93 18:40:46 IST
Date: Wed, 5 May 93 18:40:46 IST
From: nls@cse.iitb.ernet.in (N L Sarda)
Message-Id: <9305051310.AA18047@kailash.cse.iitb.ernet.in>
To: tsql@cs.arizona.edu
This is in reply to the message from Rick. His message is
reproduced below :
-----------------------------------------------------------
I think that Nandlal's classification scheme has a lot going for it.
The user's point of view has the advantage that certain types of
queries that are expected to be more frequent than others can be
emphasized.
I had a difficult time understanding the classification scheme, perhaps
in part because the explanation was necessarily short. There is at
most a sentence or two on each of the portions of the scheme.
It also is apparent that this classification scheme, while related in
some ways to Christian's proposal, is also a quite different approach
to organize queries. So my question is, is the proposed scheme (1) an
addition to, (2) a modification of, or (3) a replacement for the
existing strawman proposal?
If (1) or (2), could Nandlal modify the strawman proposal,
incorporating his recommended additions/modifications, and post a new
proposed classification scheme? If (3), could Nandlal write it up, to
the level of specificity and detail of the existing strawman taxonomy
proposal, so that the community could decide between them? Finally, it
would be very helpful if Nandlal could provide an approximate time
frame for doing so. Much work needs to be done on the step following
the taxonomy in the next six weeks, and we need to move beyond the
taxonomy very shortly. (Of course, after the workshop there will be
the opportunity to start on the second version of the benchmark,
addressing things like aggregates; at that time we could also address
improvements to the taxonomy.
-----------------------------------------------------------
I will refer to the taxonomy proposed by Christian simply as CSJ
Proposal in the following. It has the following advantages :
- it satisfies the three criteria listed by CSJ. The
classification is comprehensive (within the confines of the
decision to exclude 'group-by' and 'having' features of SQL).
- it leads to identification of orthogonal components (which are
derived from SQL features).
- It provides for suitable naming of classes and categories. User
oriented class names can be introduced here.
In my proposal, I took a different approach. I apologize for its
poor readability and inadequate explanations. I gave a syntax
"road map" for defining various classes. I tried to keep out any
SQL bias from classification (which probably is not essential).
My proposal intends the following :
- define classes which are user oriented
- identify more likely classes to check that these are more
easily expressible in TSQLs
- make classification at higher level by employing temporal
algebra level operations (rather than using only the time domain
operations in select and where clauses of SQL).
Rick's observations are well-taken. In response to the questions
posed by him, my answers are as follows :
- It is difficult to add to or modify CSJ's proposal to include
my suggestions because the approaches are quite different.
- I could attempt to add the suggested classes (in my proposal)
as certain classes/categories in CSJ proposal in Sec. 1.3, but I
have problem in making them orthogonal.
Hence, I propose to make an alternative classification scheme. I
will try to prepare a 'strawman proposal' for it. I will also re-
structure my proposal totally to improve its readability. I will
also barrow from CSJ's proposal wherever necessary and possible.
It will probably be more SQL-oriented than my earlier proposal.
Time frame : I may be able to complete the preparation of my
proposal by May 15, 1993. This may probably be too late for the
workshop. In that case, we could comment on it in future.