Conditional endorsement triple processing order "is important"

[deeglaze]

The order should not be important so long as additions are followed by triple processing that has a condition that could match it. We don’t want to impose other order dependencies unless you need to stratify triple processing more explicitly.

Given that we’re defining a processing order for base triples, it should also be allowable for profiles to specify when their triples get processed if indeed order is important. Will they always be after the base triples, or can they be specified to happen at any time?

[deeglaze]

See also #321

[nedmsmith]

Given that we’re defining a processing order for base triples, it should also be allowable for profiles to specify when their triples get processed if indeed order is important. Will they always be after the base triples, or can they be specified to happen at any time?

Evidence is the only CM type that doesn't require a condition. All the other CM types presume some sort of condition. Even the optional stages imply some sort of ACS condition has to exist before the stage can be processed.

Profiles shouldn't, IMO, be permitted to change the basic assumption that an ACS condition has to be satisfied as a pre-condition to processing a profile defined triple. The spec normative claims that the ACS in in a consistent state at the end of phase 4. If a profile is permitted to inject itself before phase 4 it should be held to the same standard. But possibly, the spec should not permit this. Essentially, we would be saying profile defined phases are always optional and hence would be applied when the other optional phases are performed. I'm not sure I have an opinion about the ordering of optional phases. Ideally, that wouldn't matter, but if it does, it would be OK for the profile to say which other optional phases should go first. At least that is my initial thinking.