On Apr 27, 2011, at 4:09 AM, Jeroen van der Ham wrote:
On 26/04/2011 16:11, Roman Ćapacz wrote:
W dniu 2011-04-22 23:40, Aaron Brown pisze:
Hey Roman,
Hi Aaron,
Overall, it looks good. A few questions on the schema.
1. It'd be good if StartTime/EndTime/Duration were specified
explicitly instead of being anyElement* (e.g. an xsd:string that
contains a unix timestamp)
I did this way to have it very generic. The UML diagram from
nml-base.pdf does not define too much details. For Lifetime it has only
"sequence of (start,end)".
I agree that we could do with a better specification of those dates and
the duration. Iirc there are some standard types in XML that can handle
that.
Looks like there are some defined in
http://www.w3schools.com/Schema/schema_dtypes_date.asp . They're kinda verbose, and don't seem to offer sub-second granularity, but I'm not sure either is all that important.
- Relatedly, is there any reason to have both Duration and EndTime
instead of just one or the other?
I think there are some scenarios where you'd use one or the other, or
perhaps even both where you ask for a link that you will use for some
duration, but it has to be before endTime.
I'm not sure the semantics all that intuitive. e.g.
<lifetime>
<duration>3600</duration>
<endTime>1304510973</endTime>
</lifetime>
Does the above mean "the circuit finishes at time 1304510973 and was active for 3600 seconds", or "the circuit is valid no later than time 1304510973, but was active for 3600 seconds sometime before then". Does this change if startTIme is specified as well?
I'm sure the US can come up with a crazier scheme. I'll write my congress critters :)
Cheers,
Aaron