New section: Treat Proxy-Scheme as having a default #43
Conversation
| For a unified view on requests, it can be beneficial for implementers (especially of servers) | ||
| to treat Proxy-Scheme the same way as Uri-Host | ||
| in that it expresses a part of the request URI, | ||
| and in that it has a default value influenced by the scheme indicated in the transport's URI compositon rules. |
There was a problem hiding this comment.
The term "default" really should be used for CoAP's Option processing only.
With a "default", proxy-scheme would always have a value and Proxy-Scheme-Number could not be introduced.
There was a problem hiding this comment.
Having a "default" helps senders and and recipients agree on a value, and helps emphasize that conceptually it is always there.
Having alternative expressions for that override does indeed make things complicated. Maybe I can come up with text that is more "Conceptually it's always there so it doesn't really do anything to set it", which would also have the nice effect of being less contradicting/updating and more explaining.
There was a problem hiding this comment.
So this really should be text that applies when neither Proxy-Scheme nor Proxy-Scheme-Number are present.
We could also put this text into href, so we don't have to put anything here.
There was a problem hiding this comment.
How can text in href influence CoAP implementations? While it's a useful tool to avoid decoding and reencoding, normatively it's just a different way of expressing a URI (reference).
There was a problem hiding this comment.
How can text in href influence CoAP implementations?
By defining the CoAP Option Proxy-Scheme-Number ....
There was a problem hiding this comment.
That would make Proxy-Scheme-Number not be a 1:1 analog for Proxy-Scheme … but if we're willing to go there (I am), yes, that could state that a present Proxy-Scheme-Number option that matches the default scheme of the transport is equivalent to its absence. That'd effectively force implementations that use it to apply the same reasoning to Proxy-Scheme as well.
If this were any other WG's document I'd protest that a new option only works out when taking a particular stance, but as href is a CoRE document, why not.
2 participants
When I brought this up in today's interim, this sounded like a good idea to add in a section.