Kernel: two smaller gaps in what it is given #39

Activity

1. don-petry

   added a parent issue on Oct 11, 2026

   • Epic: nine gaps in the Kernel, found in review #29
2. don-petry commented on Oct 14, 2026

   

   don-petry

   on Oct 14, 2026

   ContributorAuthor

   More actions

   #39 Kernel: two smaller gaps in what it is given

   Working ruling (Solution Architect, delegated). The Maintainer may overturn it in review.

   Question. Where should the facts say whether the list of Runs was read in full, and how is a configuration that names no deterministic Lens refused?

   Options.

   1. Add complete.runs to each item's facts, set by the port from its single read of the list, so the existing rule I-20 defers. Add minItems: 1 to required_lenses.deterministic in the configuration schema.
   2. Put a flag in the context for each pass. That adds a second rule beside I-20, and a second mutant.
   3. Refuse an empty set in the Kernel. That is a branch in code for something the schema can say.

   Ruling: 1.

   • One rule should cover every list that was not read in full (L-09, L-53, I-20).
   • Every ceiling is counted from Runs, so a short list must defer (I-13).
   • FR-70 requires a Lens grounded in execution. With none, I-27 and the dependency rule of Kernel: a dependency update needs no model review #32 are true of nothing.

   Acceptance criteria.

   1. complete.runs is required in the facts schema, the example, tests/support.py and sim/world.py. Counter-example: examples/invalid/facts.no-completeness-for-runs.json.
   2. test_incomplete_facts_never_converge_or_admit also covers runs: outcome defer, reason facts_incomplete. Its mutant makes incomplete() skip runs. The simulator gains a fault that cuts the list short.
   3. examples/invalid/config.no-deterministic-lens.json is rejected by tools/check_examples.py, with its test in tests/test_tools.py.
   4. Documents: design §3.3 complete; L-53 names the test.

   Confidence. High. I would rule otherwise in no foreseeable case: if the list is read once per repo instead of once per pass, the flag is still set per item.

   Needs the Maintainer? No.
3. don-petry commented on Oct 15, 2026

   

   don-petry

   on Oct 15, 2026

   ContributorAuthor

   More actions

   Note (2026-10-14): criterion 3, minItems: 1 on required_lenses.deterministic with the counter-example config.no-deterministic-lens.json, lands in "Kernel: a dependency update needs no model review" (#32). That story makes an empty list dangerous. This story keeps complete.runs.
4. don-petry

   closed this as completedin #75on Oct 15, 2026