Reacquire browser glyphs, click unsupported menu providers physically and retain exact picker diagnostics. Keep individual waits unchanged; allow the measured expanded full suite to emit its TRX within a bounded 600-second process budget.
Refs #1, #14
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Compare real PDF glyph scaling and padding, user-theme content width, optional border rows and full-page output through native controls. Verify actual paired find/source selection and screen paint after successful exports. Preserve original print cases and retain generated PDF/PNG evidence.
Refs #12
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Original hosted print cases pass: both cancels, actual default/themed PDF pixels and unchanged rendered/source/collapse state. Remove the temporary workflow and numeric probes; retain original regressions and require full clean native/published/stable/dev gates. Record measured standards readiness and remaining inventory/review/signing work.
Refs #12, #1
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
The hosted native edit reports a typed prefix while queued input is still processing. Wait for exact full readback before Save rather than asserting an incomplete snapshot; keep PDF output and reading-state assertions unchanged.
Refs #12
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Hosted evidence confirms PDF generation moves the same screen glyph by 41 pixels. Restore validated vertical geometry after the awaited PDF operation only on the same ready navigation; assert actual glyph/view rectangles instead of a stale 200-percent UIA range.
Refs #12
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
The first viewport trace reports an impossible 200 percent before printing. Pair each print stage with actual heading and browser rectangles before choosing a restoration change.
Refs #12
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Record fixture-only viewport stages and await stable native layout before taking the unchanged-state baseline. Preserve the original PDF pixel and after-print state assertions.
Refs #12
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Use the established find control identities. Exercise real filename typing with exact readback to diagnose a saved PDF missing from its expected fixture path.
Refs #12
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Keep reading and print styles in separate media sheets; preserve viewport, source, collapse, find and window font zoom. Add saved reference print choices, Windows print UI and explicit PDF save through the installed App SDK picker. Core363/100% lines, build/format/docs and55 script tests pass. Three hosted native/output cases are pending; they cancel physical print UI and rasterize owned PDFs using Windows, without a new renderer dependency.
Refs #12, #1. ADR-0003 proposed for maintainer review.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Original source viewport test passed on 690be07 in run 38121389088. Remove temporary workflow, clicked-content/focus probes and geometry traces; keep the actual browser/source/physical-wheel regression and full required suites. Preserve historical session notes and keep startup state concise.
Refs #10, #11, #13.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
WinUI ItemClick supplies the TextBlock content rather than the ListViewItem. Attach the numeric heading identity to both native elements and accept the tagged clicked content. The original physical repeat-navigation regression reproduces the mismatch; no source scroll mapping or assertion changes.
Refs #10, #11.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Refs #10, #11. Physical repetition also fails; trace the opted-in numeric ItemClick boundary and assert actual browser alignment before source alignment.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Refs #10, #11. Preserve the original glyph and wheel assertions; distinguish native ItemClick from the UIA selection path.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
The original source regression now passes actual native and browser middle-heading alignment after bounded physical wheel input. It then reproduces a same-heading click that never jumps back because selection is unchanged. Handle native ItemClick as well as SelectionChanged; preserve keyboard selection jumps and bounded numeric renderer commands.
Zero-warning build, format, docs and normal hooks pass. The existing regression is retained unchanged; hosted full/native/installed proof remains required.
Refs #9, #10, #13
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Measured input moves source and reader correctly, but a 31.8-detent event moves only one 580px viewport. Send at most four detents per event, read settled native movement after each, and retain exact glyph adjustment and all original bidirectional/wheel assertions. No application change.
Zero-warning build, format, docs and normal hooks pass. Existing 330 core tests/100% measured coverage and 55 script tests remain; hosted proof pending.
Refs #10, #13
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
The UIA range accepted min0/max7370 and value2455.883 but did not reveal the middle glyph or produce a new source geometry frame. Exercise the user gesture instead: measure one settled physical wheel detent, approach the text region, then align from actual native glyphs. Retain all source/reader alignment and wheel assertions; no application change.
330 core tests/100% measured lines retained; zero-warning build, format, documentation and hooks pass. Hosted proof remains pending.
Refs #10, #13
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
The expanded native suite passes 48/49 including actual zoom glyph/key/toolbar/reload acceptance. Preserve the original source failure and record native scrollbar min/max/read-only/value plus owned geometry before another driver change. No app change or relaxed assertions.
Refs #10, #13
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Adapt pinned reference font-size scaling, 1.1 steps and 30–500% bounds without changing line height or persisted theme JSON. Ctrl+plus/equal/minus/0 retain owned reading focus; native menu and accessible percentage toolbar share the command. Navigation retains viewport/collapse and existing find replay.
Source focus alignment passes on 1c4d1ad. Drive the actual native scrollbar after the observed no-op text-range request, retaining exact source/reader glyph and physical-wheel assertions. Separate optional diagnostics from required CI.
330 core tests/100% measured lines, zero-warning build, format, docs, all 55 script tests and normal hooks pass. Hosted 49 native/published and 48 installed tests remain pending; no local GUI input.
Refs #10, #11, #13
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Hosted diagnostics show GettingFocus captures the correct 5311 offset but later pointer/caret processing resets it to zero. Position only an unselected read-only caret in the visible text before focus; retain active find and viewport guards. Exact full UI acceptance remains required.
Allow the enlarged full published/installed suites a bounded five minutes and retain process output before reporting timeout failure. An owned headless timeout probe and both script-analyzer passes passed.
Refs #10, #11
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Keep full required UI suites intact. Run the exact failing source case on a read-only hosted Windows job and print owned focus logs before diagnostic range reads.
Refs #10, #11
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Add stable named editor choices, bounded legacy migration, copied add/remove/reorder drafts and native file/folder launch menus. Preserve one literal process argument and add actual hosted selection/persistence/probe tests. Restore the synchronized source viewport after native focus/caret revelation, with actual glyph assertions retained.
Refs #10, #11
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Physically focus the owned native source before its text-range scroll request and require actual source glyph alignment before reverse synchronization. Retain original heading bounds and physical-wheel assertions; trace actual source offset changes numerically.
Refs #10
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Hosted source selections exposed a two-character drift after two newlines. Preserve canonical offsets and normalize only the native projection to one CR per break; retain exact viewport, fold and find assertions.
Refs #10
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Avoid scanning the remaining original text for every heading on a long raw HTML line; verify 300 distinct same-line anchors. Record full native/published/installed vim acceptance while retaining full source synchronization scope.
Refs #10
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Carry parser-derived UTF-16 provenance through sanitation using fresh private markers that are stripped before presentation. Cover Setext, fences, raw HTML, nesting, duplicates, Unicode, frontmatter and forged attributes. Source scroll and collapse UI remains pending.
Refs #10
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
The first vim producer retained all 41 existing successes but UIA focus left native logical focus empty or on the outline button. Use the proven real MOVE/Down/Up path inside exact owned browser bounds, keeping real viewport, heading and query-typing assertions unchanged.
Refs #10
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Route same-document SourceChanged events through an exact-document and bounded-heading policy. Rearm repeated links after recollapse; retain physical click and actual reveal outcomes. Core 206 tests/100% lines and build/format/docs pass.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Require the physical pointer inside the exact hyperlink bounds and its owned browser provider. Retain the real click and collapsed heading/body reveal outcomes; record the proven native Ctrl+F acceptance.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Persist a validated editor executable, expose native configuration and launch commands, and preserve the viewer on missing files or launch errors. Add bounded argument tests and hosted actual-process fixtures; editor-list and folder parity remain tracked in issue #11.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Limit the Ctrl+F adapter to this foreground window and UI thread. Keep physical keyboard/link assertions and retain exact focus, pointer and fixture screenshot evidence.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Keep the physical browser-focus shortcut test and identify collapsed browser headings separately from outline labels. Correct ADR spacing and record the terminal acceptance evidence.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Declare the native source surface before the embedded browser without changing its visual column. Replace cursor-only warping with checked SendInput movement after hosted traces showed drag start/completion without deltas; retain physical mode and resize assertions and add source-divider acceptance.
Refs: #8, #9, #10
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Persist source visibility and width, refresh the committed snapshot on reload, and keep the reader usable with both panes open. Add hosted literal/read-only/reload acceptance and opt-in native event traces for the two outstanding pointer failures.
Refs: #8, #9, #10
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>
Refs #5. Preserve every configured rule and fail closed; verify both partitions with negative fixtures. All 13 files and a bounded 20-case race probe pass locally.
Co-Authored-By: Codex (GPT-6) <noreply@openai.com>
Signed-off-by: James Maes <jmaes@dmdbrands.com>