Skip to content

Restore simple cat-and-dog updates and clarify in-editor upgrades - #33

Merged
zhoy0409-debug merged 5 commits into
mainfrom
codex/simple-update-companions
Oct 18, 2026
Merged

zhoy0409-debug merged 5 commits into
mainfrom
codex/simple-update-companions

Conversation

@zhoy0409-debug

@zhoy0409-debug zhoy0409-debug commented Oct 18, 2026 •

Copy link
Copy Markdown
Owner

Restore the original tiny monochrome cat and add a matching simple dog, with compact pointer feedback and animation tied to real updater states. Clarify the first official Windows Setup installation and subsequent in-editor download, verification, saved-project confirmation, installation and reopen workflow. Portable/unregistered copies receive an explicit official Setup link.

Studio 0.1.23 / suite v1.4.23 preserve scientific, editor and native updater safety logic. The original Windows binaries remain fixed; later external reports identify the packaged build-time documentation snapshot. Two acceptance harness repairs retain real document identity and independent legacy import expectations, and preserve failed installer logs/checkpoints. Scoped QA attributes retain original JSON evidence bytes across Git checkouts.

Validation recorded in docs/qa/minimal-companions/windows-0.1.23/:

  • Actual Windows Setup install/start/frozen self-test, same-version reinstall and normal uninstall; saved project/settings/external data preserved.
  • Installed Chrome headless checks: full public Iris payload (150 observations, 12 source SVGs), edit/Undo/Save/Open, theme exports and actual cat/dog pointer feedback.
  • Selected functional coverage: 72 P0, 4 resource, 1 blank, 5 ordinary selection and 16 discovery checks.
  • Actual manual original-Setup 0.1.22→0.1.23 automatic project recovery; before/after complete saved bytes identical. Normal cleanup preserved the full external profile and finalized evidence.
  • Matched synthetic 5000-element captures retain all 46 operations per version; observed selection/recolour input→second-rAF P95: 172.9/154.6 ms on22 and115.2/135.1 ms on23. This is one-device sequential measurement, not a general acceleration claim.

The official app-driven22→23 acceptance requires the target stable assets to be public. It remains explicitly awaiting_public_stable / finalAcceptance=false in initial publication readiness. Immediately after the first stable publication, run the real editor download/helper/reopen chain and add separate evidence without replacing the original eight assets or tag. Historical failures, controlled offline metadata aborts, native IME boundaries and headless/local-device scope are documented. Exact final-head CI is required before merge.

@zhoy0409-debug
zhoy0409-debug marked this pull request as ready for review October 18, 2026 06:21
@zhoy0409-debug
zhoy0409-debug merged commit d65fb86 into main Oct 18, 2026
8 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