UWS times

[Markus Demleitner]

Hi,

On Thu, Mar 24, 2016 at 07:54:28AM -0700, Matthew Graham wrote:
> The IVOA should be consistent in its standards in how this is dealt
> with. The date used in VOEvent for the creation of the packet is
> assumed to be UTC (with a trailing Z in the example given in the
> spec) and a subset of STC is used for describing the actual event
> so the timeframe is explicit here. What other standards mention
> time?

I would wager practically all of them -- which is why DALI is a good
place for discussing at least some basics.

Registry, on the other hand, also has OAI-PMH to deal with.
Fortunately, OAI-PMH specifically defines an XSD type, UTCdatetime,
that happens to coincide with Paul's proposal (e.g., sect. 3.3.1,
"The legitimate formats are YYYY-MM-DD and YYYY-MM-DDThh:mm:ssZ.",
where the alternatives are because the protocol supports different
granularities).

In our own standards, there's the date element that's of type
xs:dateTime (which allows timezone specs, albeit restricted to Z and
literal time differences); good thing I've looked that up, because I
suspect that's more general than it needs to be.  Perhaps strong
language advising to use OAI-PMH compatible timestamp literals
exclusively would be in order.  The same is true for the @created and
@updated attributes of vr:Resource.

I'd advocate transitioning to UTC exclusively in Registry, but that's
for he future.

And given the universality of the Z I'd say we should figure out some
way to require the Z in DALI, too.

Cheers,

        Markus