[core] Re: Unsupported conditional query parameters in a request (draft-ietf-core-conditional-attributes-11)

[Christian Amsüss <[e-mail]>]

On Wed, Mar 19, 2025 at 10:04:31AM +0000, Esko Dijk wrote:
>  And in general, a CoAP server may just ignore unknown query parameters and serve the resource.)

This is not the case, the Uri-Query option is critical.

Several HTTP servers have chosen to implement things that way, to the
point where clients started relying on that behavior and now pass their
tracking information in there, but so far, this behavior has not gained
a foothold in CoAP, and I hope that it can't -- many servers I've seen
do reject unknown query parameters.

Any concrete specification is free to describe that its resource must
accept unknown query parameters, and RFC6690 and RFC9176 do that for
some of their resources, but that's up to that spec.

conditional-attributes may make any statement about `c.*` parameters,
and I appreciate that it started namespacing its parameters as to not
step onto the toes of other specs with which this might eventually be
combined, but this is something where conditional-attributes has the
choices -- from a CoAP point of view, even `/foo` and `/foo?` are
distinct, as is `/foo?bar&qux` from `/foo?qux&bar`.

BR
c

-- 
I shouldn't have written all those tank programs.
  -- Kevin Flynn