Mapping elements of VRA to a number

Question

hi,
My project involves maintaining the database for a bus company.
Each bus_route is divided into regions. Each bus-stop on a route maps to a region.
eg. bus-stop numbers 1,2,4,5,6,8 map to Region 2.
I was thinking of having a table regions with the following structure:
create table regions
Region_id NUMBER,
route_id VARCHAR2(20),
bus_stops stops_va
INSERT INTO regions VALUES(2, '459UP', STOPS_VA(1,2,4,5, 6, 8));
However, given a route_id and bus-stop number determining the region is a little tedious involving
looping through the varray elements of all rows with given route_id.
Is there an easier and more efficient way of doing this?
Amina

Answer 1

Amina
You may find it easier to stick to a relational model if you want to be able to query rapidly:
Region ---< Stop ----<Route-Stop>----- Route
Route-Stop represents the M:M relationship between Route and Stop. I'm assuming that a Stop is always in the same Region (for all the bus routes that use the stop). Assuming your bus company works like the London Underground (stop = station) then for example
Stop "Victoria Station" is in region 1;
It is on 3 lines (District, Circle, Victoria Line)
It is stop #4 on Victoria Line; (counting from Brixton)
It is stop #12 on District Line; (counting from Ealing Broadway)
It is stop #8 on Circle Line (counting anti-clockwise from Edgeware Road)
(see map at http://www.tfl.gov.uk/tfl/pdfdocs/colourmap.pdf). Actually, for timetabling purposes its on at least 10 lines (Circle line both ways, Victoria line both ways, District Line both ways to each of Wimbledon, Richmond, Ealing ...)
I'm presuming that the Route-Stop would have a stop_number (start to finish of route).
So if you want to list out a bus route from beginning to end:
select rt.route_id, rt.route_desc,
st.stop_id, st.stop_desc,
rs.stop_number,
rg.region_id, rg.region
from routes rt
inner join route_stops rs on rs.route_id = rt.route_id
inner join stops st on st.stop_id = rs.stop_id
inner join regions rg on rg.region_id = st.region_id
where rt.route_id = :my_route
order by rs.stop_number
route_stops has columns (route_id, stop_id, stop_number); the PK is all the rows together; you could make this an index organised table.
Having a relational structure makes it easy to query from any viewpoint; eg if you want to print timetables for a specific bus stop, you easily could...
Failing that, if you really want to pack the route into a nested table (and you've realised that it can be inconvenient) then it's the M:M relationship that wants to be nested. The 'route-stop' table should be nested into the routes table, not the regions table. Essentially, the stop is a more or less permanent fixture; routes may change; you can regard the sequence of stops as a (list) attribute of the route (if the route is axed, the sequence of stops is axed too). When you change a route, you change the list of stops. You'd want to edit one row in routes, not every combination of region/route. Your proposed table means that a route crossing regions would have several rows in the 'regions' table - not very well normalised :-)
HTH, but let me know if you want more information on manipulating nested tables in a relational way
Regards Nigel