home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.cs.arizona.edu
/
ftp.cs.arizona.edu.tar
/
ftp.cs.arizona.edu
/
tsql
/
doc
/
tsql.mail
/
000096_csj@iesd.auc.dk _Fri Apr 30 02:30:50 1993.msg
< prev
next >
Wrap
Internet Message Format
|
1996-01-31
|
4KB
Received: from iesd.auc.dk by optima.CS.Arizona.EDU (5.65c/15) via SMTP
id AA29105; Thu, 29 Apr 1993 17:32:40 MST
Received: from yellow.iesd.auc.dk by iesd.auc.dk with SMTP id AA07192
(5.65c8/IDA-1.5/MD for <tsql@cs.arizona.edu>); Fri, 30 Apr 1993 02:30:50 +0200
Date: Fri, 30 Apr 1993 02:30:50 +0200
From: "Christian S. Jensen" <csj@iesd.auc.dk>
Message-Id: <199304300030.AA07192@iesd.auc.dk>
To: tsql@cs.arizona.edu
Subject: TSQL Benchmark: Task 2
********************************************************************
* The TSQL Benchmark Initiative -- Task 2: Database Instance *
********************************************************************
Dear contributor, these are the benchmark tasks which must be
completed before the TDB workshop in June.
Task 1: Decide on a db schema
Task 2: Decide on an instance for the schema
Task 3: Decide on a taxonomy of benchmark queries
Task 4: Populate the benchmark with queries of each type identified in
the taxonomy.
I feel it is now time to conclude the discussion about Task 2 for the
first version of the TSQL Benchmark. To meet the deadline, we need to
move on.
A New Db Instance
*****************
I think Jim originally wanted a db instance that allowed us to
capture, with the first version of the benchmark, how grouped models
in some respects are more user-friendly than ungrouped models.
As this position is shared with several others, the db instance now
reflects this.
This could be reflected in the instance in several ways. For example,
we could introduce surrogates, or we could remove the reference to
real-world objects (e.g., persons) in the description of the instance.
The latter is my favorite. We could add two names, Ed and Edward,
which could then be interpreted as belonging to the same person or to
two distinct persons. However, Jim's specific proposal has been
adopted--it is both concrete and a straightforward modification.
Explicitly Stated Constraints
*****************************
We cannot state explicitly every constraint that the db instance
obeys. While this is true, I still feel that we should attempt to
point to *important* constraints that are satisfied by the instance.
The omission of future time as well as continuous attributes, such as
temperature, in the database should be mentioned.
Are Any User-Level Models Grouped?
**********************************
No existing user-level data models appear to be grouped. (It can still
be argued that the consensus TSQL should be designed to be grouped.)
Note that a discussion of this does not block the progress of the
benchmark!
[Aside: This claim was made by Rick 10 days ago and has not been
disputed.
I assume that in a grouped model, it is possible to express queries
like the following.
Assume a relations schema with a Name and a Dept attribute where both
Name and Dept are keys, but where persons and departments may change
names (i.e., Name and Dept are time-varying). For any instance,
(a) retrieve the times when person currently known as Ed was in the
department currently known as the Toy department, and
(b) retrieve for each department the number of persons ever to work in
the department.
I am not capable of expressing this in Gadia's TODS 88 model (which is
a very good and widely published attribute value-stamped model that I
know relatively well), and nobody has responded to this when I asked
if it was possible a couple of weeks ago. I have not yet tried with
other candidate models.]
In my next message, I will summarize the status of Task 3, with the
goal of initiating Task 4.
Best regards,
Christian