Databases Reference
In-Depth Information
We can make it more accurate by rephrasing it thus: Supplier S1 was most
recently appointed on July 1, 1999. What is more, if this version of State-
ment 1 really is what our hypothetical 2-tuple is supposed to mean, then
Statement 2 in its present form is not a fully accurate interpretation
eitherit needs to be rephrased thus: Supplier S1 was not a contracted sup-
plier on June 30, 1999, but has been one since July 1, 1999.
Observe now that Statement 1 expresses a time at which a certain event
took place, while Statements 2 and 3 express an interval in time during
which a certain state persisted. We have deliberately chosen an example in
which a certain state might be inferred from information regarding a certain
event: Since S1 was most recently appointed on July 1, 1999, that supplier
has been in the state of being under contract from that date to the present
day. Classical database technology can handle time instants (times at which
events occur) reasonably well; however, it does not handle time intervals
(periods of time during which states persist) very well at all, as we will see in
Section 5.3.
Observe next that although Statements 2 and 3 are logically equivalent,
their forms are significantly different. To be specific, the form of Statement 2
cannot be used for historical records (because since implies currency),
while that of Statement 3 canprovided we replace the phrase the present
day in that statement by some explicit date, say, September 25, 2000. (Of
course, the statement would then correspond to a 3-tuple, not a 2-tuple.)
We conclude that the concept of during is very important for historical
records, at least for state data if not for event data.
Terminology: The time(s) at which a certain event occurred or the inter-
val(s) during which a certain state persisted are sometimes referred to as valid
time. More precisely, the valid time of a proposition p is the set of times
at which p is believed to be true. It is distinguished from transaction time,
which is the set of times at which p was actually represented in the database as
being true. Valid times can be updated to reflect changing beliefs, but trans-
action times cannot; that is, transaction times are maintained entirely by
the system, and no user is allowed to change them in any way (typically, of
course, they are recorded, explicitly or implicitly, in the transaction log).
Note: The references in the foregoing paragraph to intervals and sets
of times tacitly introduce a simple but fundamental ideanamely, that an
interval with start time s and end time e actually denotes the set of all times
t such that s
e (where < means earlier than, of course). Though
obvious, this simple notion has far-reaching consequences, as we will see in
the sections to come.
£
t
£
