measurement-map misnomer? (measurement/claim/assertion/property/proposition...?)

[deeglaze]

The measurement-map type is used for non-measurement claims, such as attesting environment topology and endorsements. Shall we rename measurement-map to claims-map, and measurement-values-map to claim-values-map ?

[deeglaze]

Let's determine in this issue which term we want to use for the logical propositions that are used by endorsements, reference values, and abstract verifier state.

[deeglaze]

To recap the discussion,

• "Measurement" is not particularly helpful given that endorsements can be about unmeasured properties, such as a certifying body's assignment of a "security level".
• "Claim" is not appropriate because while analogous to the JWT/CWT term, the general definition of key/value pair is in fact too general.
• "Assertion" is a candidate. It has precedent in the Asylo attestation terminology of assertions and their dual, conditions. It does have the unfortunate connotation of a program-crashing check.
• "Statement" is a candidate, but doesn't carry the logical connotation we want to convey.
• "Proposition" is a candidate, but might have too absolute of a connotation.
• "Property" is a candidate, but is overloaded with the JavaScript notion of "property". The distance from CoRIM->CWT->JWT->JavaScript might be too short to keep the terms from being too confusing during discussion.

However way we slice it there will be tradeoffs. I think the least offensive to folks' sensibilities given text proposed by everyone at some point is "Assertion". Let's make a final decision though.

[thomas-fossati]

I wouldn't rule out Claims: I see its generality as a feature rather than a bug. This would still be my preference.

Among the other proposals, I can live with Property or Assertion.

[nedmsmith]

The challenge is that we want to use a common CDDL structure to represent information that has multiple contexts / domains of interpretation. Namely, Evidence, Reference, Endorsement. We've found it helpful to describe Reference and Endorsement as "values" while Evidence seems appropriate to use "measurement". Those of us with a TCG bias seem happy to use 'measurement' for all three cases since that has been a convention in the TCG for a long time. A pragmatic consideration when writing prose is we typically provide qualifying wording such as "reference values" or "endorsement values", and could easily follow a convention such as "evidence values". This suggests that an unqualified "values" could be used in CDDL where the CDDL applies to all three contexts. Note that the triple itself already provides "reference" and "endorsement" context, (and in the concise-evidence context); so regardless of what wording is used deeper in the CDDL, the reader is able to disambiguate.

What is wrong with using "Value" or "Values" as the value part of a named-value pairing that involves a collection of Environment claims (aka the name) and a collection of Value claims (aka the value)?

The upper level CDDL still supplies context for Reference, Endorsement, or Evidence - as before - so using generalized wording still doesn't make the structure any less descriptive.

"Assertion" is also reasonable though 9334 uses "assert", "asserted", and "asserted information" frequently to describe in more general terms "Claim" and other hand-wavey concepts. Nevertheless, it doesn't specifically use "assertion".

Nevertheless, I'm not passionate about any particular alternative, nor am I passionate about changing what's already there as I believe the CDDL supplies the needed disambiguation context. Hence, it doesn't feel like a problem that needs fixing IMHO.