[MOBY-dev] Verifying data

[Paul Gordon]

On another note, last month we talked about a "ping".  Can we enshrine 
this in the API?  The current service input doc:

http://biomoby.open-bio.org/CVS_CONTENT/moby-live/Docs/MOBY-S_API/InputMessage.html

implies you need at least one input.  Anyone mind if a I add a little 
paragraph saying how a blank mobyContent must be replied with 
mobyData-less response?  Or does someone object to enforcing this?  
Should we do an RFC intsead?  I realize it might not be backward 
compatible for a few working services, but it'll be SO useful to get rid 
of all the dead links in clients...the RDF isAlive could be based on this...

Paul Gordon wrote:
> I agree with Martin.  My MobyServlet will throw an exception if the 
> float is blank.  The stricter we are, the better off MOBY-S is in the 
> long run.  Support for broken HTML tags since Netscape 1.0 still haunts 
> the Web...
>   
>> My 2c's:
>>
>> An empty output means no output. Which may not be an error but which
>> definitely means that we have nothing to pass to a downstream service.
>>
>> Also a case when a service A returns an exception indicates that there is no
>> much sense to call a downstream service B.
>>
>> My conclusion is therefore: Each service should check first that it gets an
>> input (well, each services that expects an input, of course). If if does
>> not, it should stop and again produce no output (potentially, adding its own
>> exception to those already existing in the service notes).
>>
>> I can add this strict behaviour to Moses, or to Perl Moses.
>>
>> Cheers,
>> Martin
>>
>>   
>>     
>
> _______________________________________________
> MOBY-dev mailing list
> MOBY-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/moby-dev
>
>
>