peek
experiments in a post-browser web
Star
Fork
Watch

Commits on wip/window-kernel-executor

dietrich ayala
docs(window-manager): defect B is a wrong fact, not a missing ESC rule
committed
dietrich ayala
docs(window-manager): the kernel has three open defects again
committed
dietrich ayala
docs(signoff): record the 2026-08-21 run — three defects, one a hang
committed
dietrich ayala
docs(tests): record the third macOS gate run and its empty failing set
committed
dietrich ayala
fix(repo): add apps/server to the workspaces so its deps install
committed
dietrich ayala
docs(window-manager): answer the last open design question
committed
dietrich ayala
test(desktop): assert the new-note editor signal instead of the closing panel
committed
dietrich ayala
test(desktop): stage the second-quick-view test for the machine that runs
committed
dietrich ayala
docs: the peek window key names the slot, not the address
committed
dietrich ayala
test(desktop): the second macOS gate run settles both unattributed failures
committed
dietrich ayala
docs: the two key-namespace defects are fixed; what remains is ownership
committed
dietrich ayala
fix(desktop): a hybrid window key is private to the feature that declared it
committed
dietrich ayala
fix(desktop): a restored window keeps the source that opened it
committed
dietrich ayala
docs: window identity is the one window policy with no machine behind it
committed
dietrich ayala
fix(desktop): a peek window's identity is its slot, not its address
committed
dietrich ayala
docs: record that a peek always opens its own window
committed
dietrich ayala
docs: the ESC/child-content question is settled, not an open decision
committed
dietrich ayala
docs(desktop): record the peek-close hang fix and the macOS sign-off run
committed
dietrich ayala
fix(desktop): stop the overlay attach/hide echo race from spinning the kernel pump
committed
dietrich ayala
test(desktop): reproduce the two-page-host peek-close overlay hang
committed
dietrich ayala
fix(desktop): default the tray icon pref to off for fresh profiles
committed
dietrich ayala
docs: staged sign-off pages must be http, not data: URLs
committed
dietrich ayala
feat(desktop): stage a known starting state for the macOS sign-off run
committed
dietrich ayala
test(desktop): restage transient-autoclose exemptions on ownership, not role
committed
dietrich ayala
docs: record the first macOS gate run and what it did not settle
committed
dietrich ayala
docs: the second-quick-view failure is not Linux-only
committed
dietrich ayala
fix(desktop): only a hide of the focused window can claim the next focus
committed
dietrich ayala
fix(desktop): a lagged AppKit show no longer re-attaches the chrome
committed
dietrich ayala
test(desktop): name why the overlay is still attached, not just that it is
committed
dietrich ayala
fix(desktop): a window restored maximized can be unmaximized again
committed
dietrich ayala
docs: point the izui-escape references at the test's current name
committed
dietrich ayala
docs(window-manager): the overlay race is live on macOS too — correcting an earlier reading
committed
dietrich ayala
docs(window-manager): the restore-time maximize seed never reaches the kernel
committed
dietrich ayala
test(session-restore-page-host): cover hybrid geometry round-trip, and pin the restore-time maximize defect it found
committed
dietrich ayala
docs(window-manager): the load-bearing specs re-measured on the rebased tip
committed
dietrich ayala
test(boot-window-inventory): unmask the fixed boot-count guard, and record the overlay in it
committed
dietrich ayala
docs(window-manager): the overlay race is closed on macOS; the escape policy drifted
committed
dietrich ayala
test(izui-escape): guard the palette-opener fix on the fact it actually changed
committed
dietrich ayala
docs(window-manager): the izui-escape failure is a superseded spec, not a kernel defect
committed
dietrich ayala
test(izui-escape): the palette-opener test survives the palette's dismissal
committed
dietrich ayala
docs(window-manager): first macOS evidence — one regression open, the overlay one closed
committed
dietrich ayala
docs(desktop): a components or editor failure is not automatically a regression
committed
dietrich ayala
docs(window-manager): the sign-off covers the focus rule most likely to break on macOS
committed
dietrich ayala
docs(window-manager): the overlay spec file is a race, and two Linux failures get their cause
committed
dietrich ayala
docs(window-manager): comments name the resolvers that replaced the deleted ones
committed
dietrich ayala
docs(window-manager): record the minimal-stacking restore measurements
committed
dietrich ayala
Reapply "feat(window-kernel): stacking replays only the windows that must move"
committed
dietrich ayala
docs(window-manager): record the diagnosed cause of the overlay retarget failures
committed
dietrich ayala
fix(window-manager): stop a hide from being read as the user picking another window
committed
dietrich ayala
fix(window-kernel): the app-resign hide sweep leaves on the raw edge for every floater, not just the ones that can't hold key
committed
dietrich ayala
docs(window-manager): a second overlay publish defect gates the window-operation patterns
committed
dietrich ayala
fix(hybrid-overlay): a late subscriber is answered by its own declaration
committed
dietrich ayala
docs(window-manager): sign-off check for the raise-is-a-reorder change
committed
dietrich ayala
fix(desktop): a raise on a window already on screen reorders, never shows
committed
dietrich ayala
docs(window-manager): the raise-on-focus rule loops where the platform re-picks focus
committed
dietrich ayala
docs(window-manager): the overlay attach plan is implemented, and the stacking suppression over-claims
committed
dietrich ayala
refactor(hybrid-overlay): the machine performs the detach, so nothing outside it does
committed
dietrich ayala
refactor(hybrid-overlay): the machine decides the attach target, the overlay only paints
committed
dietrich ayala
feat(window-kernel): the machine decides which host the chrome overlay frames
committed
dietrich ayala
feat(window-kernel): a window declares that it tracks the base-layer top
committed