Open sebbader opened 3 years ago
Ideally, the whole set of HTTP methods should be supported, to provide no limitations. In any case the following methods should be at least supported:
Other methods (HEAD, CONNECT, OPTIONS, TRACE) could be relevant, but I don't have a concrete scenario where I would use these.
The other information that is relevant are the HTTP headers, for which the question is how to model these. Should they be in a sort of map structure, or should we create semantics for standard headers (Accept, Host, etc) and a map for custom headers (for instance X-B3 headers for distributed tracing)
Hi Maaren,
I agree that the HTTP verbs are the best reference for this, even though incomplete (subscribe is missing, for instance).
However, at the moment I read the message types as a combination of the operation semantics AND the target type: ResourceAvailableMessage := PUT + Resource as Target
That gives me a hard time as it will require a lot of different messages... Do you think we can find a reasonable approach here?
HEAD, OPTIONS are being used in LDP – so I like those 😉
I am not sure about a concrete use case for HEAD and OPTIONS for either ids:Resource, ids:Connector, ids:Participant or ids:Contract...
Class | Read | Create | Update | Delete | Read Metadata | Subscribe | Execute | Notify |
---|---|---|---|---|---|---|---|---|
Resource | DescriptionRequestMsg | ResourceAvailableMsg | todo | |||||
Connector | ||||||||
Participant | ||||||||
Contract |
Can we fill such a table? Would that help us?
With "we" I mean "all of us", not "myself", just to clarify things :-)
We just realized that the current Messages need to be extended to reflect more operation semantics. For instance:
Let's use this issue to collect use cases and needs, and then prepare the update for the next IM version.