Time Instant, Interval, Object modeling
hylkevds opened this issue · comments
In STAv1.1 we have:
- TM_Instant (
2018-01-01T02:00:00Z
) - TM_Interval (
2018-01-01T02:00:00Z/2018-01-01T03:00:00Z
) - TM_Object
- Either an Instant or an Interval
OData only defines TM_Instant (EDM_DateTimeOffset), not intervals.
The proposal is to model TM_Object as a complex property with a start (mandatory) and an end (optional).
Disadvantage is that even if the Observation phenomenonTime is an instant, you'll still see an object with a "start" property.
A second point is the names we use for times. Currently the proposal has:
- Datastream/phenomenonTime
- Datastream/resultTime
- Observation/phenomenonTime
- Observation/resultTime
- Observation/validTime
- Deployment/deploymentTime
- HistoricalLocation/time
- PreparationStep/time
- Sampling/samplingTime
This is rather inconsistent, since sometime just time
is used, while sometimes the name is spelled out completely (deploymentTime
, samplingTime
). For the Datastream and Observation it is obvious a shorter name can not be used, since there are multiple times, but Deployment, HistoricalLocation, PreparationStep and Sampling all only have one time.
We could either
- Use
time
for these four times - Use the full name for these four, but what name to use?
historicalLocationTime
andpreparationStepTime
are very ugly.HistoricalLocation/arrivalTime
? That would be semantically clearer.PreparationStep/executionTime
?
There is a preference for the short versions.
Also samplingLocation -> location