Marc,
> > We experienced with this pattern in the Restlet API prototypes. The
> > reactions were mixed. I would use it only if the order of
> > the calls doesn't really matter.
> >
> I like the suggestion of a builder approach for type-safe
> building of a set of metadata.
For this use case, that could be nice indeed.
> I'm OK with being a little generic in naming the setters but I think
> we should be quite explicit about the mapping to HTTP headers (when
> used with HTTP which I expect will be the majority of the time). I
> don't like the idea of different implementations using
> different HTTP headers for the same Java artifacts, that would be
> likely to cause problems.
Mostly agreed. I would just like to leave the possibility to map other
protocols to the same API semantics, even if they don't have the exact same
headers. Mapping from HTTP protocol/headers should indeed be precise in the
spec and enforced by the TCK.
Best regards,
Jerome