28 lines
1.4 KiB
Markdown
28 lines
1.4 KiB
Markdown
# State multiselect native evidence tools
|
|
|
|
Candidate tools; no native run or acceptance is asserted by their presence.
|
|
Prepare only a closed `.local/state-multiselect-native/run-1`:
|
|
|
|
```
|
|
python -B prepare.py --root CHECKOUT --runs run-1 --check-only
|
|
```
|
|
|
|
Declare any additional unchanged snapshot labels with `--extra-unchanged LABEL ...`.
|
|
Nine original history snapshots remain mandatory. Each additional snapshot must
|
|
pass every checker/provenance/time/source check and equal baseline bytes exactly.
|
|
All three distinct history byte groups, complete-Project comparisons, input-map
|
|
and executable relationships, unchanged flags, captured-window relationships,
|
|
and exact archive inventories remain required. Source count must match the
|
|
fixture and history record. Only `views/diagram_state_view.arcadia` may change.
|
|
|
|
After completed native observation and final source verification, omit
|
|
`--check-only` and provide
|
|
`--output CHECKOUT/harness/experiments/state-multiselect-20260923/native`.
|
|
Run `python -B verify.py PACKAGE` to check a written package. Existing outputs
|
|
are refused. Helpers are preserved as pinned data; they are never executed.
|
|
Referenced full build capsules/logs belong in the sibling checks package.
|
|
|
|
Offline byte/receipt consistency does not authenticate actions, establish
|
|
usability, rerun the checker, or prove native acceptance. Native action
|
|
observations and limitations require a separate reviewed record.
|