"Darcy" == Darcy Kevin (FCA) <kevin(_dot_)darcy(_at_)fcagroup(_dot_)com>
writes:
Darcy> In retrospect, the definition of the
Darcy> ??http?? and
Darcy> ??https?? schemes (i.e. RFC 7230) should
Darcy> have probably enumerated clearly which name registries were
Darcy> acceptable for those schemes, so that the following language
Darcy> from RFC 7320 (a BCP) could be invoked against any attempt by
Darcy> an app ?? Onion or anyone else -- to inject their
Darcy> own unique brand of ??specialness?? into
Darcy> the interpretation of the Authority component of their URIs:
I think all of this discussion ignores the realities of how host
software works. Onion is not injected at the browser. It's handled at
a layer that deals with connecting to names, not connecting to URIs.
Actually being responsive to these practical concerns in how people
build real-world systems is supposed to be an important part of the
IETF. I find this ongoing discussion frustrating because people are
ignoring the impact of their thoughts on the systems we actually have
before us. There have been many nice theoretical models presented.
However we have not chosen to specify APIs in the spaces under
discussion, and the APIs that are specified do not map well onto these
models. The approach that the TOR Project has chosen does map well onto
the real-world systems.