When Markdown source is opened, the document and source panes must start with equal widths, and the vertical divider must drag left/right to adjust their split.
Current implementation has a source Thumb and persisted pixel width defaulting to 320px, clamped to 160–600px. This does not satisfy the requested 50/50 opening layout; actual drag behavior needs acceptance.
Acceptance:
Opening the source pane initializes a 50% document / 50% source split of the available reading area, excluding the outline and divider. Fresh windows and reopened source panes use this opening split.
Drag the actual native divider left/right with immediate layout feedback; remove the fixed 600px ceiling and keep both panes usable within the current window width.
Window/DPI/outline resizing keeps the user-adjusted ratio within sensible minimum widths. Closing the source returns the full reading area to the document.
Show an appropriate resize cursor and provide an accessible keyboard adjustment path.
Preserve source text/selection, paired find, collapsed sections and synchronized scroll; independent document windows retain independent layout while open.
Define migration of the old SourceWidth setting without restoring the old 320px opening default. Do not impose cross-session split persistence unless agreed with the owner.
Verify actual fresh/reopen 50/50 geometry, physical divider dragging in both directions, window/DPI/outline resizing and keyboard accessibility in native/published and installed stable/dev builds. Preserve core coverage/build/style gates.
Requested by the owner for this Windows version. Related: #10 (source synchronization), #14 (windows), #1 (full application). Implementation and acceptance are pending.
Substantial AI assistance: Codex (GPT-6).
When Markdown source is opened, the document and source panes must start with equal widths, and the vertical divider must drag left/right to adjust their split.
Current implementation has a source Thumb and persisted pixel width defaulting to 320px, clamped to 160–600px. This does not satisfy the requested 50/50 opening layout; actual drag behavior needs acceptance.
Acceptance:
- Opening the source pane initializes a 50% document / 50% source split of the available reading area, excluding the outline and divider. Fresh windows and reopened source panes use this opening split.
- Drag the actual native divider left/right with immediate layout feedback; remove the fixed 600px ceiling and keep both panes usable within the current window width.
- Window/DPI/outline resizing keeps the user-adjusted ratio within sensible minimum widths. Closing the source returns the full reading area to the document.
- Show an appropriate resize cursor and provide an accessible keyboard adjustment path.
- Preserve source text/selection, paired find, collapsed sections and synchronized scroll; independent document windows retain independent layout while open.
- Define migration of the old SourceWidth setting without restoring the old 320px opening default. Do not impose cross-session split persistence unless agreed with the owner.
- Verify actual fresh/reopen 50/50 geometry, physical divider dragging in both directions, window/DPI/outline resizing and keyboard accessibility in native/published and installed stable/dev builds. Preserve core coverage/build/style gates.
Requested by the owner for this Windows version. Related: #10 (source synchronization), #14 (windows), #1 (full application). Implementation and acceptance are pending.
Substantial AI assistance: Codex (GPT-6).
Implemented locally: fresh windows and reopening source start with equal document/source columns, excluding the outline and dividers. Proportional dragging uses the actual shared space, reserves usable minimum widths and removes the old600px source cap. Focused Left/Right resize; Home restores50/50, with visible keyboard focus. The ratio belongs to the window and is intentionally not persisted; legacy SourceWidth remains readable for old settings.
21 core layout regressions cover narrow/large windows, both drag directions, both bounds and invalid measurements. The actual physical-resize case now checks equal visible columns, real drag growth, keyboard reset/resize and equal reopening instead of a saved pixel-width field.
Verified:538 core tests pass at100% measured line coverage, zero-warning Release build and format pass. Real mouse/keyboard/DPI/outline/window-resize geometry acceptance has not run. The running Debug process still has the earlier layout; all work is uncommitted/unpushed pending existing1Password signing authentication. Keep this issue open for native/published/installed acceptance.
Substantial AI assistance: Codex (GPT-6).
Implemented locally: fresh windows and reopening source start with equal document/source columns, excluding the outline and dividers. Proportional dragging uses the actual shared space, reserves usable minimum widths and removes the old600px source cap. Focused Left/Right resize; Home restores50/50, with visible keyboard focus. The ratio belongs to the window and is intentionally not persisted; legacy SourceWidth remains readable for old settings.
21 core layout regressions cover narrow/large windows, both drag directions, both bounds and invalid measurements. The actual physical-resize case now checks equal visible columns, real drag growth, keyboard reset/resize and equal reopening instead of a saved pixel-width field.
Verified:538 core tests pass at100% measured line coverage, zero-warning Release build and format pass. Real mouse/keyboard/DPI/outline/window-resize geometry acceptance has not run. The running Debug process still has the earlier layout; all work is uncommitted/unpushed pending existing1Password signing authentication. Keep this issue open for native/published/installed acceptance.
Substantial AI assistance: Codex (GPT-6).
Current native evidence is 8abbd4d/CI38168946288: full93/97 and focused10/12,zero skips. Physical source hover passes both runs; full palette passes, while focused initial Light palette records actual#1F1F1F versus expected#75628B before pointer input. Published also reproduces that initial-color variant. This is not unconditional highlighting acceptance.
Signed/DCO 9fdfc2d uses WinUI's real native Toggle menu action instead of focus-sensitive mouse fallback; the unchanged physical hover/focus/selection and exact foreground assertions remain. Source reopening/50:50 and focus-scroll alignment remain required. Exact-head CI38170170551 is running; no installed or current-head success claim. All source regressions remain in the required focused gate.
Current native evidence is `8abbd4d`/CI38168946288: full93/97 and focused10/12,zero skips. Physical source hover passes both runs; full palette passes, while focused initial Light palette records actual#1F1F1F versus expected#75628B before pointer input. Published also reproduces that initial-color variant. This is not unconditional highlighting acceptance.
Signed/DCO `9fdfc2d` uses WinUI's real native Toggle menu action instead of focus-sensitive mouse fallback; the unchanged physical hover/focus/selection and exact foreground assertions remain. Source reopening/50:50 and focus-scroll alignment remain required. Exact-head CI38170170551 is running; no installed or current-head success claim. All source regressions remain in the required focused gate.
Current signed head 04c1da8 fixes a diagnostic masking defect: the source-scroll failure reporter queried offscreen native glyph rectangles and threw E_UNEXPECTED before saving numeric focus evidence. It now saves app logs/native tree first, reports the optional observed Win32 failure and preserves the original alignment timeout. Actual geometry bounds, physical click/wheel and bidirectional assertions remain unchanged. Numeric focus and native collapse probes are opt-in and must be removed after diagnosis before release. Prior dbda6d4:96/98 native,94/98 published,17/20 focused; source-focus alignment remains red. Current Windows run: https://github.com/KofTwentyTwo/MarkdownViewr.Windows/actions/runs/38172448726.
Current signed head `04c1da8` fixes a diagnostic masking defect: the source-scroll failure reporter queried offscreen native glyph rectangles and threw E_UNEXPECTED before saving numeric focus evidence. It now saves app logs/native tree first, reports the optional observed Win32 failure and preserves the original alignment timeout. Actual geometry bounds, physical click/wheel and bidirectional assertions remain unchanged. Numeric focus and native collapse probes are opt-in and must be removed after diagnosis before release. Prior dbda6d4:96/98 native,94/98 published,17/20 focused; source-focus alignment remains red. Current Windows run: https://github.com/KofTwentyTwo/MarkdownViewr.Windows/actions/runs/38172448726.
Signed/DCO91877f7 is pushed and running at https://github.com/KofTwentyTwo/MarkdownViewr.Windows/actions/runs/38173447118. Focus probes prove the old offscreen-rectangle binary search selects character0 at scrolloffset5297.759. The fix uses actual native visible-text hit-testing before focus; physical click/wheel and exact alignment bounds remain. Nested outline DOM correctly retains child hidden indices; its test awaits the same actual hidden accessibility peer within the existing bound, including theme/reload. Local zero-warning build/format/hooks pass; actual native/distributed proof is pending. Prior04c1da8:18/20 focused,96/98 native,94/98 published,zero skips. Keep this issue open.
Signed/DCO91877f7 is pushed and running at https://github.com/KofTwentyTwo/MarkdownViewr.Windows/actions/runs/38173447118. Focus probes prove the old offscreen-rectangle binary search selects character0 at scrolloffset5297.759. The fix uses actual native visible-text hit-testing before focus; physical click/wheel and exact alignment bounds remain. Nested outline DOM correctly retains child hidden indices; its test awaits the same actual hidden accessibility peer within the existing bound, including theme/reload. Local zero-warning build/format/hooks pass; actual native/distributed proof is pending. Prior04c1da8:18/20 focused,96/98 native,94/98 published,zero skips. Keep this issue open.
Signed 9994eda is pushed; exact CI38176727468 is now running native, focused and published/installed acceptance. Previous exact6374611 is terminal: focused19/20, full native97/98 and published97/98, zero skips. Actual sustained colors/physical hover/focus and outline pass both full variants. The remaining source test advances to line490: source middle glyph aligns but the reader heading stays offscreen.
The correction uses full native content-client offscreen character points and stops adding the scroll offset twice. Original physical bidirectional assertions, deadlines and bounds are unchanged. Bounded temporary numeric probes compare anchors/targets; they must be removed before release. Local zero-warning Release, formatting and signed/DCO hooks pass. Hosted proof of the correction and installed acceptance remain pending; no stable1.0/dev publication claim.
Signed `9994eda` is pushed; exact CI38176727468 is now running native, focused and published/installed acceptance. Previous exact6374611 is terminal: focused19/20, full native97/98 and published97/98, zero skips. Actual sustained colors/physical hover/focus and outline pass both full variants. The remaining source test advances to line490: source middle glyph aligns but the reader heading stays offscreen.
The correction uses full native content-client offscreen character points and stops adding the scroll offset twice. Original physical bidirectional assertions, deadlines and bounds are unchanged. Bounded temporary numeric probes compare anchors/targets; they must be removed before release. Local zero-warning Release, formatting and signed/DCO hooks pass. Hosted proof of the correction and installed acceptance remain pending; no stable1.0/dev publication claim.
Exact9994eda focused hosted suite passes20/20, zero failures/skips, including the original actual source-focus, both-direction heading alignment and physical wheel test. Real TRX is retained under artifacts/reader-regressions-38176727468. Full native and published/installed jobs remain live; no broad installed acceptance claim yet. Temporary probe cleanup is prepared locally and will require clean exact-head verification before release.
Exact9994eda focused hosted suite passes20/20, zero failures/skips, including the original actual source-focus, both-direction heading alignment and physical wheel test. Real TRX is retained under artifacts/reader-regressions-38176727468. Full native and published/installed jobs remain live; no broad installed acceptance claim yet. Temporary probe cleanup is prepared locally and will require clean exact-head verification before release.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
When Markdown source is opened, the document and source panes must start with equal widths, and the vertical divider must drag left/right to adjust their split.
Current implementation has a source Thumb and persisted pixel width defaulting to 320px, clamped to 160–600px. This does not satisfy the requested 50/50 opening layout; actual drag behavior needs acceptance.
Acceptance:
Requested by the owner for this Windows version. Related: #10 (source synchronization), #14 (windows), #1 (full application). Implementation and acceptance are pending.
Substantial AI assistance: Codex (GPT-6).
Implemented locally: fresh windows and reopening source start with equal document/source columns, excluding the outline and dividers. Proportional dragging uses the actual shared space, reserves usable minimum widths and removes the old600px source cap. Focused Left/Right resize; Home restores50/50, with visible keyboard focus. The ratio belongs to the window and is intentionally not persisted; legacy SourceWidth remains readable for old settings.
21 core layout regressions cover narrow/large windows, both drag directions, both bounds and invalid measurements. The actual physical-resize case now checks equal visible columns, real drag growth, keyboard reset/resize and equal reopening instead of a saved pixel-width field.
Verified:538 core tests pass at100% measured line coverage, zero-warning Release build and format pass. Real mouse/keyboard/DPI/outline/window-resize geometry acceptance has not run. The running Debug process still has the earlier layout; all work is uncommitted/unpushed pending existing1Password signing authentication. Keep this issue open for native/published/installed acceptance.
Substantial AI assistance: Codex (GPT-6).
Current native evidence is
8abbd4d/CI38168946288: full93/97 and focused10/12,zero skips. Physical source hover passes both runs; full palette passes, while focused initial Light palette records actual#1F1F1F versus expected#75628B before pointer input. Published also reproduces that initial-color variant. This is not unconditional highlighting acceptance.Signed/DCO
9fdfc2duses WinUI's real native Toggle menu action instead of focus-sensitive mouse fallback; the unchanged physical hover/focus/selection and exact foreground assertions remain. Source reopening/50:50 and focus-scroll alignment remain required. Exact-head CI38170170551 is running; no installed or current-head success claim. All source regressions remain in the required focused gate.Current signed head
04c1da8fixes a diagnostic masking defect: the source-scroll failure reporter queried offscreen native glyph rectangles and threw E_UNEXPECTED before saving numeric focus evidence. It now saves app logs/native tree first, reports the optional observed Win32 failure and preserves the original alignment timeout. Actual geometry bounds, physical click/wheel and bidirectional assertions remain unchanged. Numeric focus and native collapse probes are opt-in and must be removed after diagnosis before release. Prior dbda6d4:96/98 native,94/98 published,17/20 focused; source-focus alignment remains red. Current Windows run: https://github.com/KofTwentyTwo/MarkdownViewr.Windows/actions/runs/38172448726.Signed/DCO91877f7 is pushed and running at https://github.com/KofTwentyTwo/MarkdownViewr.Windows/actions/runs/38173447118. Focus probes prove the old offscreen-rectangle binary search selects character0 at scrolloffset5297.759. The fix uses actual native visible-text hit-testing before focus; physical click/wheel and exact alignment bounds remain. Nested outline DOM correctly retains child hidden indices; its test awaits the same actual hidden accessibility peer within the existing bound, including theme/reload. Local zero-warning build/format/hooks pass; actual native/distributed proof is pending. Prior04c1da8:18/20 focused,96/98 native,94/98 published,zero skips. Keep this issue open.
Signed
9994edais pushed; exact CI38176727468 is now running native, focused and published/installed acceptance. Previous exact6374611 is terminal: focused19/20, full native97/98 and published97/98, zero skips. Actual sustained colors/physical hover/focus and outline pass both full variants. The remaining source test advances to line490: source middle glyph aligns but the reader heading stays offscreen.The correction uses full native content-client offscreen character points and stops adding the scroll offset twice. Original physical bidirectional assertions, deadlines and bounds are unchanged. Bounded temporary numeric probes compare anchors/targets; they must be removed before release. Local zero-warning Release, formatting and signed/DCO hooks pass. Hosted proof of the correction and installed acceptance remain pending; no stable1.0/dev publication claim.
Exact9994eda focused hosted suite passes20/20, zero failures/skips, including the original actual source-focus, both-direction heading alignment and physical wheel test. Real TRX is retained under artifacts/reader-regressions-38176727468. Full native and published/installed jobs remain live; no broad installed acceptance claim yet. Temporary probe cleanup is prepared locally and will require clean exact-head verification before release.