coulomb/frontend-patterns
5
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Exercise feature activation rollback and old-session artifact retention

Browse source 
Assistant: codex
Assistant-Model: gpt-6-astra
Assistant-Session: 01a0cffe-206f-7323-89d3-b20ca4cf4bf9
This commit is contained in:
tegwick 2026-09-26 23:14:34 +02:00
parent b8eec87f08
commit e34ec3fafe
22 changed files with 1661 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

18
workplans/FEP-WP-0003-contrasting-schools.md
View file

@ -150,3 +150,21 @@ host, failed load, activation rollback with fresh and already-running sessions,
shared startup outage. No production workflows/credentials needed. Navigation checks
and broader router-lifecycle limits remain live; T03 and human-validation gate unchanged.
No model calls, deployment, canonical admission or proven generativity benefit.
## Native-module publication/activation/rollback experiment — 2026-09-26
T01 progress: research/experiments/feature-release executes seven scenarios in
Chromium/Firefox, 14/14 expected observations. Compatible R2 activation changes fresh
sessions, not old document bindings; incompatible R3 evaluates before mount rejection;
404 feature failure preserves neighbor navigation; rollback affects fresh/reloaded
sessions; retiring R1 breaks old lazy sessions while loaded R1 persists; shared startup
503 blocks both actions by fixture policy. No CDN, live deployment, backend writes,
model calls or execution of upstream example workflows. Inputs/packages/browser hashes
and request traces recorded; negative controls distinguished from desirable behavior.
Independent Feature Boundary and provisional VC-06 updated, seventeen hypotheses.
Next T01 limits: old-session retention/support policy, concurrent activation updates,
safe reload with unfinished work; navigation/history/cache evidence remains pending.
Do not build a production release platform to satisfy this research model. T02/T03,
independent-human gate and M1/M2 completion remain unchanged.
See research/reviews/FeatureReleaseExperiment.md.