[rm-wg] Fwd: Lifecycle Extensiond

[David Snelling]

Begin forwarded message:

> From: David Snelling <David.Snelling at uk.fujitsu.com>
> Date: 14 March 2007 22:48:40 GMT
> To: "Strong, Paul" <pstrong at ebay.com>
> Cc: David Snelling <David.Snelling at UK.Fujitsu.com>
> Subject: Re: Lifecycle Extensiond
>
> Paul,
>
> On 6 Mar 2007, at 18:34, Strong, Paul wrote:
>
>> Hi Dave,
>>
>> So I've looked at these and have been thinking...
>>
>> I like the mind map idea a lot as a way of viewing our problem  
>> space in
>> a qualitative sense, although I am typically somewhat averse to the
>> notion of transition states such as Commissioning and so forth.   
>> Where
>> do you stop? :o)
>
> I'm not that sure either. WRT management, the next bit of  
> information a manager might need is "what is supposed to happen  
> next to this thing?" That may be what we were after.
>
>> I would view them as sub-states of the existing
>> states.
>
> For the *-ing states that are intended as optional sub-states.
>
>> We define the test that defines what it means to be in a given
>> state.  Those tests can be very simple.  Below are some ideas but  
>> they
>> need discussion/-fleshing out and I'm pretty sure that the Configured
>> test is not right yet. ;o)  But you get the idea.
>>
>> Active State - Test = Fulfils its function - i.e. is responding to
>> transactions or is still doing its calculation.  Sub states include
>> running, stopping, failing (before it has ceased to serve its  
>> purpose),
>> canceling etc.
>
> Looks OK.
>
>> Configured - Test = Has a full set of configuration and data files
>> required to run, even if those may cause it to fail.  The point is  
>> that
>> the component can be started.  Sub-states include startable,  
>> starting,
>> failing (the bit after it no longer satisfies the serving purpose  
>> test),
>> stopping, canceling, decommissioning.
>
> I still like Commissioned and separating out the "bit after it no  
> longer satisfies the serving purpose test" states.
>
>> Discovered - Test = Has been registered by the system, i.e. has a  
>> unique
>> identifier and associated records of its properties and  
>> attributes.  Sub
>> states include deleting, configuring.
>
> The test i right. For me, it has a WS-Addressing EPR and will  
> respond to WSRF-Lifetime actions. One it is WSRF-LT:Destroyed, it  
> ain't there non more. My adjective, "Extant", was recommended by my  
> kids, based on the first question the family always asks in the  
> "I'm thinking of an animal." game. - "Is it extinct or extant?"
>
>> We can also perhaps sub-class (so to speak) the transitions to  
>> include,
>> for example, Stop(Failed), Stop(Succeeded) and so forth if that is
>> possible, otherwise we can perhaps add transitions.  It's all  
>> about how
>> we define the states and the transitions.  I might posit the
>> sub-transition notion to Grady if I can't find a suitable extant
>> mechanism.
>
> We did a lot of brainstorming with UML 2.0 guards etc to try and  
> capture all this, but the thinking process get hard to verify. Sub- 
> typing the transitions, if we keep them in the data structure,  
> might work. the problem we get in the end, is this all has to be  
> rendered in XML comprehensible to fairly brain dead tooling.
>
>> I think the discussion is really useful.  And as you know I am a  
>> massive
>> fan of simplicity at the abstract level that can then be built  
>> upon, so
>> that complex entities and behaviors can be represented yet easily
>> understood.
>
> I like this flavor ice cream too, but I like a sauce on it that  
> says, how far can we push the abstraction before it breaks. Then  
> back off a little and try to simplify again. I think this is a  
> healthy tension.  See you at OGSA.
>
>
>
> -- 
>
> Take care:
>
>     Dr. David Snelling < David . Snelling . UK . Fujitsu . com >
>     Fujitsu Laboratories of Europe Limited
>     Hayes Park Central
>     Hayes End Road
>     Hayes, Middlesex  UB4 8FE
>     Reg. No. 4153469
>
>     +44-208-606-4649 (Office)
>     +44-7768-807526  (Mobile)
>
>
>

-- 

Take care:

     Dr. David Snelling < David . Snelling . UK . Fujitsu . com >
     Fujitsu Laboratories of Europe Limited
     Hayes Park Central
     Hayes End Road
     Hayes, Middlesex  UB4 8FE
     Reg. No. 4153469

     +44-208-606-4649 (Office)
     +44-7768-807526  (Mobile)