Request.scheme is not set properly or at all about californium HOT 9 CLOSED

My understanding is that we are talking about an incoming CoAP request, right?
The problem here seems to be that (currently) Californium does not know whether the incoming request has been sent via coap or coaps. The problem is that Californium's CoAP layer is independent from the underlying transport layer so we will need to figure out a way for the CoAP layer to determine whether the Endpoint we have received the request on is secure (DTLS) or not.

from californium.

Yes, you are right. Currently it is not possible to know if a request is coming from a coaps endpoint. So if a CoAP peer wants to send a follow up request to a peer, using the RDNodeResource.getContext(), sends in this new request as plain (non-DTLS) CoAP request to a port listening for DTLS CoAP requests.

In Class CoapEndpoint.InboxImpl where request object created, is this a valid fix?

if (raw.getCorrelationContext().get(DtlsCorrelationContext.KEY_SESSION_ID) != null) {
    request.setScheme(CoAP.COAP_SECURE_URI_SCHEME);
} else {
    request.setScheme(CoAP.COAP_URI_SCHEME);
}
request.setSource(raw.getAddress());
request.setSourcePort(raw.getPort());
request.setSenderIdentity(raw.getSenderIdentity());

from californium.

I guess additionally, RDNodeResource class need to be changed to take the request.getScheme() into consideration to properly set the RDNodeResource.context variable.

from californium.

@calif-devel,

In Class CoapEndpoint.InboxImpl where request object created, is this a valid fix?

I had exactly the same idea :-)
However, I think we should encapsulate this test in RawData, e.g.

if (raw.isSecure()) {
    request.setScheme(CoAP.COAP_SECURE_URI_SCHEME);
} else {
    request.setScheme(CoAP.COAP_URI_SCHEME);
}

This way we do not expose too much of the underlying mechanism around CorrelationContext to upper layers ...

Would you like to create a PR? :-)

from californium.

I like the fix.

Another comment: Do not rely too much on the RD implementation. It was a contribution that has never been fully overhauled, and hence is of lower code quality... (yet I did some improvements a while ago).

This means, whenever you find something strange in the RD code, do not hesitate to ask about it! ;)

from californium.

Fixed and submitted as Pull Request #42

from californium.

I will take a look (and merge) tomorrow.
Thanks again, @calif-devel

from californium.

Hi @calif-devel,

I guess you can now close this issue ... thanks again for contributing :-)

from californium.

Thank you @sophokles73 and @mkovatsc for resolving this.

from californium.