I think another way of saying this is that OpenSocial doesn't depend on "views", but that doesn't mean a given implementation *can't* include it as an implicit dependency. When it happens this is a confusing situation, so it's one we should probably discourage or try to remove where it occurs. It leads to gadgets that aren't spec-compliant yet still work in certain cases.
As a quick pedantic follow-up, I'd suggest in the future moving these sorts of questions to the OpenSocial and Gadgets Spec discussion group in our ongoing effort to keep implementation and spec separate ( http://groups.google.com/group/opensocial-and-gadgets-spec). Thanks, John On Mon, Mar 31, 2008 at 3:13 PM, Kevin Brown <[EMAIL PROTECTED]> wrote: > Absolutely not. Please read the spec: > http://code.google.com/apis/gadgets/docs/spec.html > > An implementation of opensocial that happened to depend on views might be > OK > in some cases (say, for implementing requestShareApp), but that fact > should > be well hidden from the gadget developers. > > There are too many transitive dependencies as is. It makes upgrading > libraries independently of one another a godawful nightmare, and evolving > the spec impossible. > > On Mon, Mar 31, 2008 at 2:02 PM, Dan Lester <[EMAIL PROTECTED]> wrote: > > > > > I've come across a couple of gadgets that use the 'views' feature to > > work out if they're on the canvas or profile. > > > > However, these do not explicitly require 'views' in their spec (only > > 'opensocial'). They work OK in Orkut et al but not in Shindig. > > > > Should Shindig's opensocial feature include a dependency on views? > > > > Sorry if I've missed the point. Perhaps the other containers are just > > trying to be extra helpful... > > > > Dan > > > > > > > -- > ~Kevin >
