Modules whose BEHAVIOUR the gate has asserted, with what was asserted.
Under wrangler dev an enable killed the host process, so no follow-up request could be made and
nothing could be verified. Re-run under @cloudflare/vitest-pool-workers that limit does not
exist: an enable survives, a follow-up request answers, and TWO enables in one object survive --
the exact case that killed wrangler dev hardest. The failure was miniflare's proxy controller, a
component that only exists locally, and suspecting the instrument first was right.
What that leaves is a configuration gap rather than a runtime one:
admin_toolbar and ctools are verified, with what was asserted recorded above.
pathauto enables cleanly and has NOTHING TO APPLY: no pathauto.pattern.* config ships, so
a node save produces no alias. Installed, ok: true, and silently inert.
token is unverifiable for the same reason. Its behaviour is token replacement, and the only
observable on this site would have been the alias pathauto did not generate.
A deploy would not change any of this: the blocker is absent config, not the memory envelope, so
it would reproduce identically on real infrastructure.
Modules whose BEHAVIOUR the gate has asserted, with what was asserted.
Under
wrangler devan enable killed the host process, so no follow-up request could be made and nothing could be verified. Re-run under@cloudflare/vitest-pool-workersthat limit does not exist: an enable survives, a follow-up request answers, and TWO enables in one object survive -- the exact case that killed wrangler dev hardest. The failure was miniflare's proxy controller, a component that only exists locally, and suspecting the instrument first was right.What that leaves is a configuration gap rather than a runtime one:
admin_toolbarandctoolsare verified, with what was asserted recorded above.pathautoenables cleanly and has NOTHING TO APPLY: nopathauto.pattern.*config ships, so a node save produces no alias. Installed,ok: true, and silently inert.tokenis unverifiable for the same reason. Its behaviour is token replacement, and the only observable on this site would have been the alias pathauto did not generate.A deploy would not change any of this: the blocker is absent config, not the memory envelope, so it would reproduce identically on real infrastructure.