Re: [Cbor] Packed CBOR review

[Christian Amsüss <[e-mail]>]

Hello CBOR group,

we've just spent some time discussing the use of packed CBOR in CoRAL,
and the current working assumptionis that we'll use the very same
mechanism.

That means we'll do some practical exploration with something similar to
both setup formats that are currently around:

> Spiffy-Standard-A-Table-Setup = #6.4711(any)
>
> Table-Setup-Uri = #6.52([Uri-reference, any])

and see how it applies there.

One thing we'll probably keep of our older designs is the bounded
import: Previously we described our data source tables as lists of
(source, first-import, last-import) where the latter two are numeric and
allow picking just a slice of the external dictionary. This has two
benefits:

* It keeps the number of (efficient) low-number imports slow -- if you
  need two of here and two of there, you can express all of them in
  single bytes even though both dictionaries may contain 20 items each.

* It provides extensibility. That's even practical on non-extensible
  dicts (as a consumer might make sense of the document not knowing how
  many items are in there), and more so with extensible dicts (which of
  course can only be extensible by adding more entries but not by
  mutating earlier ones, that'd be a recipe for desaster).

Hoping for a frugal exchange of ideas and experiments
Christian

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