[MOBY-dev] Primitives as primary input

Tom Oinn meredith at cantab.net
Thu Jan 10 19:09:47 UTC 2008


Mark Wilkinson wrote:
> I agree with Paul (or did Paul agree with me??) that email and url are 
> namespaces.  Paul's argument that it doesn't make sense to sub-class an 
> email or a url is a compelling one, which also gives me some assurance 
> that this decision is correct.

There are plenty of cases where you might want to restrict the permitted 
semantics of a URL, most of which boil down to a restriction on what the 
URL can point *to*.

Email addresses are different, in that they all have the same inherent 
content. They're not really references in the same way as URLs.

It's somewhat broken in my view to have URL in moby *at all*, it would 
be better to have a framework where you could explicitly seperate the 
reference from that to which it refers. If you have a tool that accepts 
a URL to an image the signature of the tool is 'image', the URL part of 
that is the delivery mechanism and in an ideal world they wouldn't be 
conflated in this way. The exception would be something that took a URL 
as a string to, e.g. validate it according to the URL specification (in 
which case the type is 'url' and the carrier type is 'string value').

If you could define types in terms of their carrier mechanism (URL, SCP 
path, value etc) and split this out from the actual content they carry 
(image, person, bio-foo-object) this would make life a lot easier (and 
is, in a completely non-coincidental fashion, what Taverna2 does for 
exactly this reason).

Of course that's just a 'small' change to how Moby works :)

Tom



More information about the MOBY-dev mailing list