[MOBY-dev] {Spam?} Re: The Agent behaviour - discussion and questions to the MOBY community

[Paul Gordon]

Reading my own e-mail, I have an idea:  make the empty-ping behaviour 
part of the MOBY API (which clients MAY call), AND require a valid test 
case in the RDF description (which MOBY Central WILL call to measure 
robustness).  The best of both world, no?
> Sorry, I should have expressed this more clearly. I definitely agree 
> with you that if you don't provide a signature URL, you shouldn't be 
> deregistered.  But, if you did provide one, you must maintain it.
>   
>>> 1. The RDF is a contract of sorts.  If you are not committed enough to
>>> keep the RDF available for download, your service should be
>>> deregistered.
>>>     
>>>       
>> Paul, you are not expressing yourself precisely. I agree that the RDF is a
>> contract. But if I do not sign to such contract (by not providing an RDF
>> signature URL) I should not be kick off (but my service should be considered
>> "testing"). I was very happy that we (at least me and Mark) seemed to have
>> consensus on this. Are you telling the same, or are you disagree (and
>> opening this can of worms again)?
>>   
>>     
> This is where we disagree.  As I said earlier, we have an opportunity to 
> enforce robustness in the protocol, and I think we should take it.  
> Obviously, the test case the service provider lists can be a fairly 
> simple one, it is up to them.  Which is more useful to the biologist, 
> knowing a service is reachable (empty ping), or knowing it works (test 
> case)?
>
> The appropriateness of an empty ping vs. a functional ping probably 
> boils down to where we expect the "robustness score" to be store, by 
> MOBY Centrals, or does each client instance store its own?
>   
>>> 2. I think the ping should be mandatory, with a valid input test case.
>>>     
>>>       
>> The valid input test cases are nice - and we can use them: we have all the
>> machinery in place. Which is: an LSID resolution service to get test input
>> data from a service provider, the predicate name for that, and
>> MOBY_Environment project that can use this test data to check services (of
>> course, you can use your own checking calls).
>>    But: it is beyond what a simple ping should do. Simple ping should not do
>> that much and it will still be useful to get fast and cheap reply if a
>> service (or at least its server) is up and running.
>>    So I am just for the ping described by Pieter. Nothing more. (And I have
>> it implementd in Perl Moses already.)
>>
>>
>> Martin
>>
>>
>>   
>>     
>
> _______________________________________________
> MOBY-dev mailing list
> MOBY-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/moby-dev
>