[Oslc-Automation] Availability Spec review - my feedback - OSLC Actions - thread 5 Availability Condition (AC)

[John Arwe]

TF you had this under AR in your response, but in the original it's under 
AC


MP> mttf & mttr: What type of resource? 
TF> I wasn't sure if I should use an integer or any other time-related 
datatype for this.... ... That's why I thought I'll define it as 
AnyResource.

AnyResource does not answer the question even for an integer, it just 
inserts 1 level of indirection.  Now, for an integer, it's probably a 
duration so it has units if it's expressed as an integer.  Now it's at 
least 2 properties (value and units), so making it a resource looks more 
obvious.  One option would be to re-use what we did in OSLC Performance 
Monitoring (re-used the Estimation and Measurements' vocabulary to express 
value + units) and make that resource an ems:Measure (or :Metric?).  If we 
go with the "likely to be an ems:"whatever language now, we can always 
switch it out later if needed (by adding a second alternative) if new 
scenarios drive the need for something more complex.  Can you say: Open 
World Assumption (thread 3) ?  ;-)

To really make your heads hurt, think of my "what does read-only=true 
mean?" response in thread 1 ... the type is part of the resource shape, 
not the vocabulary (Core has discussed whether vocabulary should constrain 
it or not).  So it's an expectation, not a guarantee [waves to Tim 
cheekily].  I've run into exactly one case so far in which I was actually 
glad for this - it saved a breaking change - although I cannot for the 
life of me remember the particulars.




Best Regards, John

Voice US 845-435-9470  BluePages
Cloud and Smarter Infrastructure OSLC Lead
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://open-services.net/pipermail/oslc-automation_open-services.net/attachments/20140514/1a9f22e8/attachment-0003.html>