From: Tom O. <tm...@eb...> - 2006-07-07 12:10:40
|
Stian Soiland wrote: > Jun Zhao wrote: > >> Currently it is fine to attach some data typing to the input or output of a *workflow*. I tried to hack the taverna UI to attach the ontology tree to the input and output of a processor. But it is not clear to me where this semantic metadata is to be kept? I don't see any places in the current Scufl. > > There's the "Edit template" option for processor that nobody knows how > work, which gives a highly confusing interface with some drop-downs for > ports. I know how it works, but it was a couple of hours of hack to see if we could introduce service specific provenance generation and I don't think anyone uses it now. > The xscufl is also not really the right place to attach such > information, as the information is general to the service, and not to > the workflow. This is where FETA comes into the picture. Actually that's not the case, there are certain kinds of metadata which are intrinsic to the service (which should be stored in a registry ideally) and some which are contextual, they only apply to that use of the service. This is especially true for generic operations such as the set and string manipulation ones, in a given workflow they have a specific meaning over and above their generic one. Tom |