Re: Bidirectional Binary Self-Joins
Date: Fri, 30 Mar 2007 19:25:50 -0600
Message-ID: <m3648iurtt.fsf_at_garlic.com>
paul c <toledobythesea_at_oohay.ac> writes:
> Not trying to change the example, which I think is excellent, but this
> reminds me of the airline game. In flight and related systems, it is
> enough to use airline code, flight number (an invented number), date,
> scheduled departure and arrival time in hours and minutes, departure
> airport code and destination airport code to indicate a flight (or
> flight segment, the term that biz uses). One could ask what prevents
> two different flights with that recorded key, from taking off from the
> same large multi-runway airport in the same minute? The answer is the
> key itself, which is embedded in every manual and computer method the
> controllers, airport personnel and so forth use, which doesn't include
> a runway code, nor anything close to scheduled times measured in
> seconds.
actually there are flights that do take off simultaneously. i blaim
it on TWA ... the first time I remember seeing it was very early 70s
... twa had plane parked overnight in san jose ... first thing in the
morning, it took off for SFO, with two different flight numbers. in
san fran, some of the passengers ... who thot they had a "direct"
flight with no connections ... found that while their flight number
went direct, they had a "change of equipment" (NOT a connection) ...
since the original equipment continued outbound from SFO with one of
the flight numbers ... and some totally different equipment assumed
the other flight numbers for a different destination.
in a different life, we were given the opportunity to rewrite ROUTES
... one of the common airline res system applications ... i.e. finding
flights/times/etc to get from origin to a destination. as part of the
effort, we were provided a machine readable copy of the OAG ... with
all world-wide commercial flight segments. I believe the "worst" I
found was what appeared to be six different flight segments
... i.e. different flight numbers but identical equipment, identical
departure times from the same airport and identical arrival times at
the same destination airport.
when i expressed my opinion about the "change of equipment" scenario
... the explanation was that agent reservation screens (as well as
printed manuals) typically ordered all direct flights first on the
screen (or in books) before all connecting flights. judicious use of
multiple flights numbers for the same equipment, got a lot of things
moved up to the top of the screen (with people who were avoiding
connecting flights found themselves faced with "change of equipment")
the other "benefit" (of creating multiple different flight nos for the
same equipment), was typical reservation system provided agents with
only a limited number of connecting flight operations. more complex
travel scenarios required agents to manually stitch together some
number of connections. use of multiple flight numbers per equipment
... could provide some additional trip planning help to agents.
so one of the "ten impossible" things (current ROUTES couldn't do and
we were suppose to implement), was being able to find connections
between any two airports (some 4k plus) in the world. for demo, they
would give two airports codes ... frequently ones that nobody had ever
heard of before on opposite sides of the world. there were some that
took more than 24hrs elapsed time with 5-6 different connections.
being able to automatically find any possible origin/destination
... at least eliminated the excuse (for multiple flight nos per
equipment) that the agent was being helped on how to get from any
possible origin to any possible destination.