Skip to content

Preserve finite evidence for ambiguous install output - #238

Merged
patrickchin merged 1 commit into
mainfrom
codex/android-install-ambiguity-evidence
Nov 15, 2026
Merged

patrickchin merged 1 commit into
mainfrom
codex/android-install-ambiguity-evidence

Conversation

@patrickchin

Copy link
Copy Markdown
Owner

The signing diagnostic currently reports only ambiguous-content when bounded install output contains conflicting public markers. Preserve that decline and record its first predicate, fixed count buckets, success-marker boolean, a sorted set drawn only from the existing49 public codes, and an unknown-code boolean. Observed codes never become an accepted installCode while ambiguity remains.

Unknown getter/proxy/inherited diagnostic objects fail closed. Supported native own-data result classification, the49-code grammar, status/size/path/full-frame gates, install commands, capabilities, assertions and cleanup stay unchanged. Validation: source self-check and adversarial privacy cases;5200 old/new native-result fixtures (4692 ambiguous,392 accepted);642 unchanged command-helper fixtures;9 unchanged-install assertion cases;6 canonical evidence tests; exact source/call/cleanup preservation. Author and root probes used Node26.9; this new source still needs the hosted Node22 inline predevice check. No current install cause/mode, native transition, provider or release acceptance is claimed by this PR.

@cloudflare-workers-and-pages

Copy link
Copy Markdown

Deploying puptime with  Cloudflare Pages  Cloudflare Pages

Latest commit: 1973b44
Status: ✅  Deploy successful!
Preview URL: https://9fe1f45b.puptime.pages.dev
Branch Preview URL: https://codex-android-install-ambigu.puptime.pages.dev

View logs

@patrickchin
patrickchin merged commit ae4cb1d into main Nov 15, 2026
5 of 6 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant