diff options
| author | Teddy Wing | 2017-05-30 15:14:45 +0200 |
|---|---|---|
| committer | Teddy Wing | 2017-05-31 17:57:36 +0200 |
| commit | 43c14bc8c916d00a9c478f1e78d5860977c56bbe (patch) | |
| tree | 7d65e70711c10bfc75303bb963a130f7b2cb5d9f /lib/sql/clone_schema.sql | |
| parent | c6d622d662d7f787787d8ee3d89f5d900dd8ca59 (diff) | |
| download | chouette-core-43c14bc8c916d00a9c478f1e78d5860977c56bbe.tar.bz2 | |
VehicleJourneyAtStopsDayOffset: Calculate day offsets
Add a rough implementation that determines the day offsets of a sequence
of VehicleJourneyAtStops.
It currently passes the first (and only test) associated with it.
My first go at it assumed that the dates meant something, but really
they're just there for decoration. All the dates will be on '2000-01-01'
regardless of their day offset. This is because they need to be stored
as dates, so the application has standardised on a single placeholder
date. That go didn't work because the after-midnight times were still on
'2000-01-01'.
The new version bases the day offset on the times. It does this by
checking whether the current time is earlier in the day than a previous
time. If it is, we assume that the day has changed. This works when the
gap between times is never 24 hours or more.
It will not work if, for example, we have a "current day" time at 10:00
and a "next day" time at 10:00. We have no way to infer that they're on
different days.
This implementation appears to work, at least for the case the test is
checking, but I don't like it because it doesn't read well. Hoping to
change it in addition to adding my other test cases.
Refs #3596
Diffstat (limited to 'lib/sql/clone_schema.sql')
0 files changed, 0 insertions, 0 deletions
