The scheduler comparison is not automatic

The scheduler package $\mathsf{S}$ records acyclic dependency, finite-experiment termination, commutation of independent operations, schedule-independent complete records, sensitivity equal to strict reachability in its designated models, and confluence. It is not automatically defined for every process matrix CheungXXIIA,CheungXXIIB.

On a declared common model universe $\mathfrak{U}$, let $C$ denote operative class restrictions and $H$ the additional assumptions. The required additional-content audit is

separate from

Where the baseline is not applicable, its comparison dictionary must be supplied instead of assigning $\mathsf{S}$ by name.

En08-A records the scheduler obstruction and its ball-section and orthant realizations CheungEn08A (thm:main). For the standard orthant in dimension at least three, homogeneity holds through positive diagonal automorphisms, but strict convexity fails because $e_1+e_2$ is on the boundary. For the standard diamond cone

the sum of $(1,1,0)$ and $(1,0,1)$ is again on the boundary. These standard cones fail (K3) CheungEn08A (prop:orthant, prop:ball).

Those computations are not a translation of the predecessor models into the frozen component carrier or a new selector. A relevant countermodel requires its process, access, carrier, and rule to be interpreted first. Moreover,

A Euclidean-ball section is not itself a non-Lorentzian cone. Excluding named examples is not classification of every remaining possibility.