[MOBY-dev] Primitives as primary input

[Paul Gordon]

Any takers?

Paul Gordon wrote:
> Hi all,
>
> I believe, for reasons that I've outlined before, that namespace 
> inheritance is generally a very bad idea.  I can see what Tom and 
> Andreas are getting in terms of practicality of protocol + content 
> type specification, but strongly stand by the fact that the 
> protocol/namespace equivalence should be the mechanism in Moby.  Now 
> the question is how to combine the namespace and the data type in a 
> service signature if that service is a retrieval service.  The 
> VirtualPNGImage method is currently the only way, but the Virtual idea 
> is usable for any object.  So we'd need to be able to add an attribute 
> saying "I'm virtual".  e.g.
>
> <PNGImage namespace="http" id="//google.com/images.foo.png" 
> virtual="yes"/>
>
> And we'd have to add to the service signature a mechanism for saying 
> "I'll accept a PNGImage, or a virtual one that is in the http, ftp or 
> scp namespaces".  Of course, by default, a service will only accept 
> real objects as they do know.  You'd have explicitly register the fact 
> that you take virtual data.
>
> My $0.02
>
> Paul
> _______________________________________________
> MOBY-dev mailing list
> MOBY-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/moby-dev
>
> !DSPAM:60005,478787ef10950176725398!
>
>
>