[Suit] URIs, IRIs and CRIs in SUIT manifests

[Christian Amsüss <[e-mail]>]

Hello Brendan, SUIT,

todays's meeting had the question of URI references vs. IRI references
in the manifest.

(and now Dave has chimed in advocating for URIs on the protocol level
due to specification trouble with IRIs, and his point is valid, so maybe
it is better to just stick with URIs).


From a specification PoV, I'm OK with either. URIs are a bit easier to
use in a deterministic form, but references are not expected to be
compared anyway, only resolved.

From an internationalization PoV, both URIs and IRIs allow non-ASCII URI
components -- it's just that the IRI form is human readable (and more
compact by a factor of 3 for the non-ASCII parts).

From an implementation PoV, frankly I can't tell what is supported
better: the URI processors I have encountered in embedded space are
often so limited that I'd expect handling errors in non-ASCII URIs no
matter whether they are encoded as URIs or IRIs.

The easiest to process on an embedded device would be CRIs[1] -- but
that so far has little implementation base, and is only due for
completion in May. It has advantages on the compactness side, and UTF-8
is transported without any percent encoding.

BR
Christian

[1]: https://datatracker.ietf.org/doc/html/draft-ietf-core-href

-- 
To use raw power is to make yourself infinitely vulnerable to greater powers.
  -- Bene Gesserit axiom