oficina / docs Public

Download this file

1	# Native Docs review record
2
3	2026-09-13: independent specification review dispatched as gpt-6-astra with high reasoning. Actual runtime identity was not independently verified. The amended specification was approved after adding native known-schema editing parity, twelve locales/accessibility, sandboxed CRDT decoding, bounded per-document recovery, explicit UTF-16 positions, memory-only content and current stream authorization.
4
5	Final independent production branch review: approved at `c2e1a92`, against
6	inherited `d5ff04e17144583620ff3b9950fcf6722e7001e9`, with the evidence and manual
7	acceptance limits below. The review covered the working web inventory, native
8	Mac/iOS implementation, current authorization, sealed server storage, lossless
9	CRDT identity/formatting, own-only undo, generation/recovery transitions, native
10	clipboard/export, and the three narrowly scoped Yrs deletion backports. No
11	additional concrete production blocker was established after the persisted-rule
12	regression passed. The reviewer authored the isolated UI targets/runner and
13	independent wire/deletion probes; production implementation was reviewed
14	independently. This does not claim exhaustive manual parity acceptance.
15
16	The final original server CI and native engine/Apple gate passed. All six
17	parallel Mac/simulator/device builds passed in
18	`../native-build-logs/20260913T095810.093604Z/results.json`. Source review preserves
19	normal Yrs garbage collection; it does not adopt an unbounded tombstone policy.
20
21	## Isolated native review
22
23	The review app, UI test runner, and Go fixture are separate test targets. They
24	use a fresh temporary database and the real native approval/exchange flow.
25	They never access a production account.
26
27	Run from the repository root with a booted simulator and the Go toolchain
28	required by `go.mod` on PATH:
29
30	```sh
31	python3 native/scripts/review-ios.py --simulator SIMULATOR_UUID
32	```
33
34	The runner builds the review app and UI tests before starting the fixture, then
35	installs both bundles. It starts `TestNativeReviewFixture`, checks that its exact
36	native credential can read the document list, and writes mode-0600
37	`Documents/review.json` into each current simulator container. The app and test
38	runner each read their own file; neither uses a hardcoded host temporary path.
39
40	UI execution has a seven-minute limit inside the fixture's ten-minute lifetime.
41	On success, failure, or timeout, cleanup stops the test processes, signals the
42	fixture to finish, terminates the review bundles, and deletes credentials from
43	both original and current containers. Resolving containers again handles Xcode
44	reinstalling an application during the test.
45
46	Logs and the `.xcresult` bundle remain under `native/build/review-TIMESTAMP`.
47	`--go /path/to/go` selects a toolchain explicitly. `--skip-build` reuses already
48	rebuilt artifacts. `--smoke` verifies authentication, credential installation,
49	and cleanup without claiming editor acceptance.
50
51	The orchestration smoke test passed on 2026-09-13 with simulator
52	`5064842A-8777-44F3-9AFF-E6562CF88F71`. On the same iPhone simulator, the native
53	create/type/Bold/append flow passed with exact body and formatting verified via
54	a fresh authenticated Markdown export. Its screenshot was visually inspected.
55	The runner removed both credentials and stopped the fixture afterward.
56	The same accepted binary passed the identical flow and fresh server read on
57	iPad Pro 13-inch (M5), iOS 26.5, with simulator
58	`C12B219D-0114-4EB6-A75B-6F6565D422CD`.
59
60	`--only testCreateTypeFormatAndPersistDocument` isolates that case;
61	`--only testExportPDFAndMarkdownDestinationsAndCancellation` isolates export.
62	Export acceptance opens both native destination pickers and cancels; the runner
63	asserts no private export staging remains and cleans it on failure too.
64	An actual external destination write is outside these cases.
65	On the iPad simulator, both export picker/cancel flows passed in 24.458 seconds.
66	Their screenshots were visually inspected: each picker displays “Native review
67	document”. The runner confirmed that no export staging directory remained and
68	removed both review credentials before stopping the fixture.
69
70	The latest iPhone build also passed `testCreateTypeFormatAndPersistDocument`
71	with exact saved body and bold formatting in
72	`native/build/review-20260913T092049.526154Z/Review.xcresult`; credentials and
73	fixture cleanup passed. This covers the generation-reuse and explicit-selection
74	changes as well as ordinary typing.
75
76	Additional review cases are `testInputRuleReversalAndMultilinePersistence`,
77	`testKeyboardJumpSearchOpensDocument`, and `testJumpButtonSearchOpensDocument`.
78	The final rebuilt iPhone rule case passed in 16.859 seconds in
79	`native/build/review-20260913T095637.456828Z/Review.xcresult`. It verifies native
80	prefix removal, Backspace restoration, multiline typing, and the exact fresh
81	server Markdown body. The earlier extra empty list is absent after the three
82	targeted Yrs deletion backports. Its screenshot was visually inspected; fixture
83	and credential cleanup passed.
84
85	The iPhone jump-button case passed in
86	`native/build/review-20260913T094927.493167Z/Review.xcresult`: native search uses
87	the real ranked endpoint and the selected result opens the second document.
88	The review-only `testRegisteredCommandAndSoftwareSubmitOpenDocument` case and
89	selected XCTest suite passed in 20.223 seconds, recorded in
90	`native/build/review-20260913T100847.314877Z/ui.log`.
91	It locates the actual focused native editor, verifies its registered Command-K
92	binding and priority, dispatches that fixed selector through UIKit's responder
93	chain, searches the real endpoint, and submits with software-keyboard newline.
94	The second document's native editor opens. Xcode then hung while finalizing its
95	result bundle; the owned runner was interrupted and its normal cleanup removed
96	credentials and stopped the fixture. Direct synthetic screenshots were retained.
97	This is passing XCTest evidence, not a successful final xcodebuild exit. The
98	fixed-command probe is confined to the separate review target and has no
99	production launch hook.
100
101	Physical hardware keyboard acceptance remains unverified. iPad `typeKey`
102	occasionally opened search but also inserted a literal K; hardware Return
103	injection did not open the result even after adding an explicit key handler.
104	One hardware-connection attempt had the same Return result, and restored the
105	original `ConnectHardwareKeyboard=0` preference afterward. These failed HID
106	cases remain diagnostics and are not claimed as passing. The phone keyboard
107	case explicitly skips: Apple's [XCUIElement overview](https://developer.apple.com/documentation/xcuiautomation/xcuielement)
108	documents keyboard interaction for macOS and iPadOS; attempted iPhone Command-K
109	injection produced a literal K.
110
111	Other unexecuted manual matrices remain listed in `PARITY.md`: full formatting
112	and clipboard UI combinations, simultaneous live editing across devices, native
113	sharing, VoiceOver/native-speaker review, final destination writes/print, and
114	interactive Mac close/menu behavior. Their implemented paths and targeted
115	tests were reviewed; those matrices are not represented as manually completed.
116
117	A failed XCTest case now gets 45 seconds for Xcode to finalize its result bundle;
118	if Xcode remains stuck, the runner stops only its own test process group and
119	cleans up. This failure path was exercised. The seven-minute total test bound
120	still applies. Synthetic jump screenshots are also written by the separate test
121	runner and copied into the private review log directory during cleanup, so they
122	survive an incomplete `.xcresult`; their originals are removed afterward.
123