What the named domain does and does not establish

The canonical-domain definition permits the expression $m^{\circ }$ to be stated with its correct domain restriction. It also permits the restriction map $r_O$ in Proposition prop:descent to be named. It does not provide the support relation required to regard every contact value as an element of

That relation would require the actual lift or pushforward of the collision supports into $W$ and their position relative to $O$.

In particular, the collision class cannot be inserted into $\mathrm{Hom}({\mathcal{P}}_{\mathrm{int}},Y)$ merely because its members are called contacts. A contact on one graph-coordinate collision can have a nontrivial image under a graph map. Whether that image lies outside the canonical domain is an additional geometric statement.

D0.8 is therefore patched as a named domain with an explicit non-equation. This removes the mistaken requirement that the refused equation itself must be recovered. It does not claim an operationally complete new description of every admissible configuration.

The source stop is unchanged. Canonical agreement is an agreement where the subsequent operations are already defined. It is not the characteristic prescription missing from those operations.