Platform.crash kills the daemon abruptly. lifecycle_hidden.crash turns a
monocle on, crashes, starts the daemon again, and checks that it manages
every window, that none is left out of sight, and that a monocle works
again. The KDE extra parking keeps the stale parking desktop checks.
Platform.fullscreen and Platform.user_maximize take a window fullscreen or
maximize it as the user would, without asking the daemon. The scenarios
maximizing.user and fullscreen check that the daemon takes the window out
of the layout, leaves it alone, and puts it back. The KDE extras keep the
on-top screenshot, the daemon's log and the setup.
resize.edge_cases makes its drags from the daemon's layout and reads the
frames from the operating system: both sides of a shared edge, a horizontal
edge, a corner and an edge of the screen, with a neighbour or without. The
KDE extra keeps pointer_is_exact.
A display whose mode changes: the daemon's monitor follows, the windows
stay on a display and apart, and changed back they are where they were.
Platform.resizable_display says which display can change its mode, or None
and the scenario is skipped; KDE says None, as the virtual outputs of the
headless KWin have one mode. Platform.resize_display changes the mode.
hotplug.user_move moves a window with Platform.move_window to the second
display and checks that the daemon has it on that monitor and every window
is tiled on its display. The KDE suite runs it before the unplug and after
the replug, and the extra sent_by_the_user is gone.
monocle.focus_behaviour checks that with no-op the window keeps the work
area when the focus is moved in a direction, and with cycle another window
gets it. The KDE extra monocle_focus_behaviour is gone.
The shared rules scenario sends a window by the name of a workspace, checks
that a rule that is not initial is applied again when a window opens, and
that the daemon's global state has no rule after clear-all-workspace-rules.
The KDE extras keep the desktop numbers.
The shared placement scenario opens a window with a class and a rule that
puts the windows of that class first. The KDE extra by_app_id is gone;
by_program (foot.exe) stays.
Each komorebic query is compared with the daemon's state, before and after
a change of focus and layout, and a resize that was saved is undone by
loading it, with quick-save-resize and with save-resize. Platform gets
temp_path for the file of save-resize.
commands.layouts and layouts.run close windows from the end and open them
again with their own titles. The KDE extra layouts_by_count keeps only what
reads the daemon's log: KWin has each window where the daemon last asked.
minimize, unmanage and manage on the focused window now run through the
neutral platform. The check that the unmanaged window is centered is
neutral: inside the work area and not on a tile of the layout.
Platform.spawn takes rename_to and after: the window gives itself another
title after some seconds (a terminal escape on KDE). The scenario ignores a
title, and checks that the window is left alone while it has it, and tiled
once it has another.
Starts the daemon with the layout_flip option of a workspace set to each value
it takes, over columns, rows and BSP with two and three windows. Checks the
mirror from the OS, that flip-layout turns the option off and on, that a command
adds the other axis, and that the option is the named workspace's own.
core-e2e runs it after startup_rules.
Checks from the OS and the daemon's state what send-to-monitor, move-to-monitor,
their cycle, workspace and named forms, focus-monitor, cycle-monitor,
focus-monitor-workspace and cycle-move-workspace-to-monitor do: the window on
the monitor and workspace asked for, tiled in that display, the others where
they were, and the focus where each command leaves it. focus-monitor-at-cursor
is its own function, for a platform with a pointer.
core-e2e starts KWin with two outputs and unplugs the second before the daemon,
so the other scenarios still see one display. The fake desk (two displays of
different sizes) and its tests are new files.
The audit behind the docstrings of Platform: what each member promises
to both ports, what each port answers for another reason, what each
finding of the review turned out to be, and what is left to run on a
KDE VM and on Windows.
spawn raised nowhere on KDE when the window never came, so a check
that a window stays unmanaged passed for a window that was not there.
It does raise now, as do close, minimize, activate and move_window for
a window that does not exist, and a program name Windows rejects. A
title is looked for as it is and not as a pattern. displays() is what
KWin says of each output and not the size the session was given
(Kde.screens, the same script that gives the work area).
The docstrings of Platform say what each member promises and what a
scenario may not lean on.
The quiet period of three seconds was written in nine places, each
a number. It is support.QUIET. The window the lock scenario moves is
a third into the work area, not at 400,300, and the scenario now
checks that it moved, or the check that it is back in its tile after
the unlock proves nothing. The room the padding must change by, the
pointer drag and the tile hotplug checks after an unplug come from
the work area.
The scenario toggled with --implementation windows on both ports, and
its settle time said it was KWin's 200 ms. The default of the command
is each port's own implementation, so the flag goes, and the settle is
a named wait that is longer than what either port needs.
Platform.hidden is false for a window the system does not have, so the
workspace scenarios that counted the windows that are not hidden took
a window that was lost for one that was shown. They count the windows
the system has (support.in_view). The docstring says what each port
answers true for beyond a window the daemon put away, and that a
scenario that asks for one to be put away asks the daemon too.
The decrease was checked against the frame within Platform.tolerance,
which is 24 px for a terminal: a daemon that gave back less than the
increase passed. The daemon's layout, which no client rounds, now has
to show the width asked for, and the frame has to follow within the
tolerance. Platform.tolerance says what it is and is not a bound on.
The headless suites have no panel, so the whole output was right there,
but a session with one got a work area that scenarios could not hold a
window to. KWin's placement area of the first output is asked through a
script, with the whole output as the answer where KWin does not say.
The contract of Platform.work_area is spelled out: the display less what
the system keeps for itself, as the system says. Monocle and maximize
check that the window stays inside it. The fake platform takes a panel,
and a daemon that tiles under it.
libz-sys before 1.1.31 logs the compiler command of its zlib probe, which
with cc 1.6.0 or later includes every environment variable of the build,
and cargo stores that output under target/. It is a build dependency of
shadow-rs via git2 and libgit2-sys; bump it in the lockfile.
Displays were matched on serial_number_id.eq(..) || device_id.eq(..), so
two displays with no serial number matched each other through None == None,
and two with no device path through the shared "UNKNOWN" device id. When
a display without either was unplugged, the one that stayed attached
"matched" it, nothing was removed, and every display change started the
3s verification again; state kept listing the display for good.
A serial number now identifies a display only when both have one and it is
not empty, and a display without a device path is told apart by its name.
The removal takes the monitors out by index instead of by id, so that two
displays sharing the id UNKNOWN are not removed together, and adding a
connected monitor no longer skips a display for sharing that id.
PlacementTarget deserialized through String, so the integer the docs
promise ("initial_window_placement_rules": 2) made the daemon exit at
start with an untagged enum error while "2" worked. Accept integers as
well as strings.
The derived schema described an externally tagged object that neither
form matches, so describe the type by hand and regenerate both schemas.
move_container_to_monitor and move_workspace_to_monitor took the container
or workspace out of its place and only then looked the target monitor up,
so a monitor index that does not exist left the window on screen but in no
workspace. Check the index first so the command fails and nothing moves.
The socket handlers for manage, ignore, object-name-change, tray, layered
and title bar rules skipped a new rule when any simple rule had the same
id, whatever its kind, so ignore-rule exe foo after ignore-rule class foo
was dropped. Share one helper that compares the kind as well.
Everything but dependencies.json is stored with LF, which nothing said, so
a CRLF file could get in unnoticed (it did, for two files, in a round trip
that shows as a rewrite of every line of both).
* .gitattributes: normalize every text file to LF in the repository and
check it out with LF, whatever core.autocrlf is (the Windows runner
defaults it to true)
* rustfmt.toml: newline_style = "Unix", so cargo fmt --check fails on
a Rust file with CRLF
* CI: fail when a file is stored with CRLF
* dependencies.json: the one file that was stored with CRLF