2005-11-18 Elijah Newren <newren@gmail.com>
Merge of all the changes on the constraints_experiments branch.
This is just a summary, to get the full ChangeLog of those
changes (approx. 2000 lines):
cvs -q -z3 update -Pd -r constraints_experiments
cvs -q -z3 diff -pu -r CONSTRAINTS_EXPERIMENTS_BRANCHPOINT ChangeLog
Bugs fixed:
unfiled - constraints.c is overly complicated[1]
unfiled - constraints.c is not robust when all constraints
cannot simultaneously be met (constraints need to be
prioritized)
unfiled - keep-titlebar-onscreen constraint is decoration
unaware (since get_outermost_onscreen_positions()
forgets to include decorations)
unfiled - keyboard snap-moving and snap-resizing snap to hidden
edges
109553 - gravity w/ simultaneous move & resize doesn't work
113601 - maximize vertical and horizontal should toggle and be
constrained
122196 - windows show up under vertical panels
122670 - jerky/random resizing of window via keyboard[2]
124582 - keyboard and mouse snap-resizing and snap-moving
erroneously moves the window multidimensionally
136307 - don't allow apps to resize themselves off the screen
(*cough* filechooser *cough*)
142016, 143784 - windows should not span multiple xineramas
unless placed there by the user
143145 - clamp new windows to screensize and force them
onscreen, if they'll fit
144126 - Handle pathological strut lists sanely[3]
149867 - fixed aspect ratio windows are difficult to resize[4]
152898 - make screen edges consistent; allow easy slamming of
windows into the left, right, and bottom edges of the
screen too.
154706 - bouncing weirdness at screen edge with keyboard moving
or resizing
156699 - avoid struts when placing windows, if possible (nasty
a11y blocker)
302456 - dragging offscreen too restrictive
304857 - wireframe moving off the top of the screen is misleading
308521 - make uni-directional resizing easier with
alt-middle-drag and prevent the occasional super
annoying resize-the-wrong-side(s) behavior
312007 - snap-resize moves windows with a minimum size
constraint
312104 - resizing the top of a window can cause the bottom to
grow
319351 - don't instantly snap on mouse-move-snapping, remove
braindeadedness of having order of releasing shift and
releasing button press matter so much
[1] fixed in my opinion, anyway.
[2] Actually, it's not totally fixed--it's just annoying
instead of almost completely unusable. Matthias had a
suggestion that may fix the remainder of the problems (see
http://tinyurl.com/bwzuu).
[3] This bug was originally about not-quite-so-pathological
cases but was left open for the worse cases. The code from
the branch handles the remainder of the cases mentioned in
this bug.
[4] Actually, although it's far better there's still some minor
issues left: a slight drift that's only noticeable after
lots of resizing, and potential problems with partially
onscreen constraints due to not clearing any
fixed_directions flags (aspect ratio windows get resized in
both directions and thus aren't fixed in one of them)
New feature:
81704 - edge resistance for user move and resize operations;
in particular 3 different kinds of resistance are
implemented:
Pixel-Distance: window movement is resisted when it
aligns with an edge unless the movement is greater than
a threshold number of pixels
Timeout: window movement past an edge is prevented until
a certain amount of time has elapsed during the
operation since the first request to move it past that
edge
Keyboard-Buildup: when moving or resizing with the
keyboard, once a window is aligned with a certain edge
it cannot move past until the correct direction has
been pressed enough times (e.g. 2 or 3 times)
Major changes:
- constraints.c has been rewritten; very few lines of code from
the old version remain. There is a comment near the top of
the function explaining the basics of how the new framework
works. A more detailed explanation can be found in
doc/how-constraints-works.txt
- edge-resistance.[ch] are new files implementing edge-resistance.
- boxes.[ch] are new files containing low-level error-prone
functions used heavily in constraints.c and edge-resistance.c,
among various places throughout the code. testboxes.c
contains a thorough testsuite for the boxes.[ch] functions
compiled into a program, testboxes.
- meta_window_move_resize_internal() *must* be told the gravity
of the associated operation (if it's just a move operation,
the gravity will be ignored, but for resize and move+resize
the correct value is needed)
- the craziness of different values that
meta_window_move_resize_internal() accepts has been documented
in a large comment at the beginning of the function. It may
be possible to clean this up some, but until then things will
remain as they were before--caller beware.
- screen and xinerama usable areas (i.e. places not covered by
e.g. panels) are cached in the workspace now, as are the
screen and xinerama edges. These get updated with the
workarea in src/workspace.c:ensure_work_areas_validated()
2005-07-31 Elijah Newren <newren@gmail.com>
* src/stack.c (is_focused_foreach, get_standalone_layer): use only
the expected_focus_window instead of both the focused_window and
the expected_focus_window. Removes an infinite flicker loop in
sloppy and mouse focus, and an ugly one time flicker in click to
focus. Fixes#311400.
2005-01-28 Elijah Newren <newren@gmail.com>
Patch from RHEL-3 (Havoc doesn't remember how it got there) that
Havoc posted in bug 156511 to fix the problem with fullscreen
windows on a different xinerama monitor not staying on top. I
updated to HEAD. Should fix#156511.
* src/stack.c: (windows_on_different_xinerama): new function,
(get_standalone_layer): let windows on a different screen than the
one with the focus window stay in the fullscreen layer
2005-01-27 Elijah Newren <newren@gmail.com>
Treat splashscreens same as other windows for stacking. Fixes
#165243.
* src/stack.h: (MetaStackLayer enum): remove META_LAYER_SPLASH
from the list
* src/stack.c: (get_standalone_layer): remove the special casing
of META_WINDOW_SPLASHSCREEN
2005-01-23 Elijah Newren <newren@gmail.com>
Restore original stacking when aborting an alt-esc window switch
operation. Fixes#123576.
* src/display.c: (GRAB_OP_IS_WINDOW_SWITCH): new macro,
(meta_display_close): clear grab_old_window_stacking if non-NULL,
(event_callback): restore stack positions if alt-esc op cancelled
with button press, (meta_display_begin_grab_op): store the old
stacking positions, (meta_display_end_grab_op): free the old stack
positions
* src/display.h: (struct _MetaDisplay): add a
grab_old_window_stacking list
* src/keybindings.c: (process_tab_grab): restore stack positions
if alt-esc op cancelled with an errant key press
* src/stack.c: (compare_just_window_stack_position): new
GCompareFunc function, (meta_stack_get_positions): get current
stack positions, (compare_pointers): new GCompareFunc function,
(lists_contain_same_windows): simple utility func to see if two
lists contains the same windows, (meta_stack_set_positions): new
function to set the positions of all the windows in the stack
* src/stack.h: (meta_stack_get_postions,
meta_stack_set_positions): new functions
2004-06-17 Elijah Newren <newren@math.utah.edu>
Add support for _NET_WM_USER_TIME
* src/display.c:
(meta_display_open): Add _NET_WM_USER_TIME to atom_names[],
(event_callback): Manually set _NET_WM_USER_TIME upon KeyPress
(doesn't work since keyboard isn't grabbed) and ButtonPress (does
work), this is just a fallback for applications that don't update
this themselves.
* src/display.h: (struct _MetaDisplay): Add atom_net_wm_user_time field
* src/screen.c: (meta_screen_apply_startup_properties): Check for
TIMESTAMP provided from startup sequence as well.
* src/stack.c:
s/meta_window_set_stack_position/meta_window_set_stack_position_no_sync/,
(meta_window_set_stack_position): New function which calls the
meta_window_set_stack_position_no_sync function followed immediately
by calling meta_stack_sync_to_server.
* src/window-props.c:
(init_net_wm_user_time), (reload_net_wm_user_time): new functions,
(reload_wm_hints): also load atom_net_wm_user_time
* src/window.c:
new XSERVER_TIME_IS_LATER macro (accounts for timestamp wraparound),
(meta_window_new_with_attrs): add timestamp attributes,
(window_takes_focus_on_map): use TIMESTAMP from startup
notification and _NET_WM_USER_TIME to decide whether to focus new
windows,
(meta_window_show): if app doesn't take focus on map, place it
just below the focused window in the stack
(process_property_notify): check for changes to _NET_WM_USRE_TIME,
(meta_window_stack_just_below): new function
* src/window.h:
(_MetaWindow struct): new fields for initial_timestamp,
initial_timestamp_set, net_wm_user_time_set, and net_wm_user_time,
(meta_window_stack_just_below): new function
2004-03-01 Rob Adams <readams@readams.net>
* src/stack.c (compute_layer): don't promote due to transiency; we
handle that elsewhere now.
(ensure_above): perform layer promotion here as well as stack
position promotion. Note that this means that we need to do stack
constraints now on layer change now.
(get_maximum_layer_of_ancestor): remove function
(max_layer_func): remove function
(MaxLayerData): remove struct
2004-02-27 Rob Adams <readams@readams.net>
Handle layer promotion of transient descendants of layer-promoted
windows to also be layer promoted, using a simple iterative
algorithm.
* src/stack.c (compute_layer): change name to promote_layer, and
convert to simply perform any necessary layer promotion without
computing the standalone layer.
(max_layer_func): use window->layer instead of
get_standalone_layer
(get_maximum_layer_of_ancestor): use window->layer instead of
get_standalone_layer
(meta_stack_ensure_sorted): implement iterative algorithm,
explained in a long comment.
* src/window.h: add a tmp_layer field used by stack.c for
determining if the stack is dirty or not, since maintaining this
information in meta_stack_ensure_sorted is no longer practical.
2003-08-19 Rob Adams <robadams@uclu.edu>
* src/stack.c (get_standalone_layer): put windows with
wm_state_below at the bottom. Make this higher priority than full
screen layer; see #120238.
2003-07-27 Rob Adams <robadams@ucla.edu>
* src/window.c (update_move): Update window shaking loose so that
the window is moved to the pointer and certain drag state is
properly restored once windows "reattach". Fix for #115000 based
on the patch by Jurg Billeter.
* src/screen.c (meta_screen_resize): Invalidate work areas after
an xrandr screen size update. Fix for #117230.
* src/stack.c (window_is_fullscreen_size): Check the bottom corner
of the window in addition to the top corner. Fix for #118194.
* src/constraints.c (meta_window_constrain): Support aspect ratio
hints in the new constraints code. Fix for #113798.
* src/tools/metacity-window-demo.c (toggle_aspect_ratio): toggle
the aspect ratio hints to force a 16:9 aspect ratio.
(do_appwindow): add a button to toggle aspect ratio.
2003-05-29 Rob Adams <robadams@ucla.edu>
* src/main.c (main): chdir to the user's home directory on
startup. See #113755.
* src/stack.c (get_standalone_layer): a window should be in the
fullscreen layer if it or any of its transient descendants are
focused or expecting the focus and it is either fullscreen or
fullscreen sized. Fix for #104369.
* src/stack.c (is_focused_foreach): foreach used by
get_standalone_layer to find focused transient descendants.
2003-03-11 Havoc Pennington <hp@pobox.com>
Should fix#108108, #106217, tracked down by Owen Taylor and
Frederic Crozat
* src/window.c (meta_window_foreach_transient): change
MetaWindowForeachFunc to return a boolean for whether to continue
(meta_window_foreach_ancestor): new function
(window_should_be_showing): use meta_window_foreach_ancestor
(unminimize_window_and_all_transient_parents): ditto
(update_sm_hints): ditto
(meta_window_is_ancestor_of_transient): ditto
* src/stack.c (get_maximum_layer_of_ancestor): use
meta_window_foreach_ancestor
2003-02-23 Havoc Pennington <hp@pobox.com>
Patch from Rob Adams addresses #95014 (placement issues),
makes first fit algorithm "center tile", adds most code
for per-xinerama workspaces (#86682) but disables it for now.
* src/workspace.c (meta_workspace_get_work_area_for_xinerama)
(meta_workspace_get_work_area_all_xineramas): new xinerama
functions, maintain workspace->work_areas with a different
work area for each xinerama. However for now all the work
areas are the same, because haven't quite figured out how
_NET_WM_STRUT is supposed to work
* src/window.c: adapt to new meta_window_* xinerama APIs
(meta_window_get_work_area_current_xinerama): new xinerama
API
(meta_window_get_work_area_for_xinerama): new xinerama API
(constrain_position): be a bit more clever about which xinerama's
work area we choose to use.
* src/stack.c: adapt to new Xinerama API
* src/screen.c (reload_xinerama_infos): invalidate all work areas
(meta_screen_get_xinerama_for_rect): new function
(meta_screen_window_intersects_xinerama): new function
* src/place.c (find_first_fit): change to use
"center tiling" (center a screen full of tiled windows,
rather than aligning them top left). Adapt to new
xinerama functions.
2003-02-20 Havoc Pennington <hp@redhat.com>
* src/stack.c (create_constraints): don't create constraints
between windows on different screens, #106086 tracked down
by Arvind
2003-01-25 Havoc Pennington <hp@pobox.com>
* src/stack.c (window_is_fullscreen_size): When checking if a
window is fullscreen size, only require it to be at the origin
of the work area, not at the origin of the screen/xinerama.
Still require it to be full screen in width x height.
May fix xine in the case where the user has a top panel.
* src/window.c (constrain_position): restore the ability for
undecorated windows to position themselves overlapping the top
panel, but don't let decorated windows do so. Oh the hacks...
2003-01-21 Havoc Pennington <hp@pobox.com>
Fix for the "mangles focus window when switching workspaces
and using mouse focus" bug
* src/stack.c (meta_stack_get_default_focus_window_at_point): new
function
* src/screen.c (meta_screen_focus_mouse_window): new function
(meta_screen_focus_default_window): new function
* src/workspace.c (meta_workspace_activate): use the
new meta_screen_focus_default_window()
2003-01-17 Havoc Pennington <hp@pobox.com>
* src/window.c (meta_window_handle_mouse_grab_op_event): fix event
compression code to use GDK algorithm suggested by Owen, should be
more efficient.
2002-11-12 Havoc Pennington <hp@redhat.com>
* src/stack.c (meta_stack_get_default_focus_window): never use a
window with input = FALSE take_focus = FALSE as the default focus
window #95454 fix from Hidetoshi Tajima
2002-11-03 Havoc Pennington <hp@pobox.com>
* src/window.c (meta_window_new): use window-props.h stuff for a
couple of properties
(implement_showing): fix printf string
* src/xprops.c (meta_prop_free_values): new function
* src/window-props.h, src/window-props.c: start moving code that
handles loading window properties into this file.
2002-10-26 Havoc Pennington <hp@pobox.com>
* src/iconcache.c: include config.h
* src/group.c: include config.h
* src/frame.c: include config.h
* src/core.c: include config.h so it doesn't crash all over the
place due to #ifdef HAVE_STARTUP_NOTIFICATION
* src/util.c (meta_print_backtrace): export from this file
* src/main.c (log_handler): print backtrace here
2002-10-21 Havoc Pennington <hp@redhat.com>
Optimizations for managing new windows (do not all take effect if
METACITY_DEBUG=1). Bug #96404
* src/keybindings.c (meta_change_keygrab): use error trap nesting
and conditionalize on meta_is_verbose() to avoid a ton of XSync
* src/display.c (meta_change_button_grab): ditto
Throughout: move to new error trap setup to save on XSync calls,
new setup is:
* src/errors.c (meta_error_trap_push_with_return): new function,
an error trap that needs to care about return value and thus
sync even if an outer trap still exists
(meta_error_trap_pop_with_return): new function
(meta_error_trap_pop): add "last_request_was_roundtrip"
argument allowing us to avoid XSync() if we just did
a GetProperty or whatever.
* src/util.c (meta_warning): flush the warning file descriptor
* src/Makefile.am (INCLUDES): define G_LOG_DOMAIN
2002-10-18 Havoc Pennington <hp@pobox.com>
* src/stack.c (constrain_stacking): replace the old
apply_constraints with wacky new approach involving graphing all
the constraints then walking the graph. Fixes#94876 and probably
other stacking bugs as well, thanks to Arvind for tracking down
the issue.
(compute_layer): add FIXME and reference to bug #96140
2002-10-17 Havoc Pennington <hp@redhat.com>
* src/stack.c (apply_constraints): don't place
transient-for-whole-group windows above _each other_, only
above other windows in the group that aren't themselves
transient-for-whole-group. Should help with part of #94876
2002-10-17 Havoc Pennington <hp@redhat.com>
* src/stack.c (apply_constraints): fix memory leak of
group_windows, and don't use the variable name "tmp" twice. Shadow
variables bad.
2002-10-11 Havoc Pennington <hp@pobox.com>
* src/stack.c (window_is_fullscreen_size): make the checks here
allow windows larger than the screen in addition to
exactly-screen-size
* src/window.c (meta_window_configure_request): delete the "try to
auto-enter-fullscreen-state" hack here, because it was broken, and
the changes to the stacking code to move screen-size focused
windows to the fullscreen layer should work better.
(meta_window_new): remove auto-fullscreen hack from here too
2002-10-09 Havoc Pennington <hp@pobox.com>
* src/stack.c (apply_constraints): also keep utility/menu/toolbar
windows above their whole group.
(get_standalone_layer): don't use META_LAYER_FOCUSED_WINDOW, but
only use META_LAYER_FULLSCREEN while the fullscreen window has
focus. Also, put screen-sized windows in the fullscreen layer,
even if we didn't dare to actually put them in the fullscreen
state.
2002-10-01 Havoc Pennington <hp@pobox.com>
* src/screen.c (meta_screen_resize_func): make it static
* src/stack.c (get_standalone_layer): put above/below windows
in an appropriate layer.
* src/screen.c (set_supported_hint): say we support above/below
* src/display.h (struct _MetaDisplay): add _NET_WM_STATE_ABOVE,
_NET_WM_STATE_BELOW atoms
* src/window.c (meta_window_client_message): handle above/below
state messages
(set_net_wm_state): handler above/below state
(update_net_wm_state): handle above/below states
2002-09-28 Havoc Pennington <hp@pobox.com>
* src/window.c, src/stack.c: Rewrite stack code to work a lot
differently. Should be better now, and not lose relative positions
of windows when moving among layers. Also should now be possible
to get session management to restore the stacking order. Probably
breaks some stuff, and makes all the stack.c changes I made
yesterday sort of irrelevant.
2002-09-27 Havoc Pennington <hp@pobox.com>
* src/stack.c (get_standalone_layer): Temporarily disable use of
the FOCUSED_WINDOW layer, because given the fact that moving
multiple windows into the same layer changes the Z-order of those
windows, it was breaking click-to-focus.
2002-09-27 Havoc Pennington <hp@pobox.com>
* src/screen.c (meta_screen_focus_top_window): raise the focused
window, since it may not be the window on top, given the below
change.
* src/stack.c (meta_stack_get_default_focus_window): make this
more complex to prefer to focus the transient parent, followed by
other windows in group, followed by topmost non-dock, followed by
dock. Previously was just topmost non-dock followed by dock
ignoring groups and transiency.
2002-09-27 Havoc Pennington <hp@pobox.com>
* src/place.c (constrain_placement): constrain placement to try to
keep windows from going offscreen to the right/bottom
* src/stack.c (compute_layer): rearrange the logic here to say
that a window must always be in at least as high a layer as any of
its transient parents or group members, rather than special-casing
fullscreen. Also, group_member_is_fullscreen was leaking the list
of group members every time, a fairly major memory leak.
2002-09-24 Havoc Pennington <hp@redhat.com>
* src/window.c (meta_window_update_layer): new function
* src/stack.c (compute_layer): put focused window in a layer above
all other windows, in click-to-focus mode. #93022
* src/window.c (meta_window_notify_focus): update window layer on
focus change.
2002-08-24 Havoc Pennington <hp@redhat.com>
* src/util.c: translate some strings that should have been, and
convert to locale encoding before printing stuff.
* src/stack.c (group_member_is_fullscreen): if window itself is
fullscreen, return TRUE immediately.
* src/window.c (meta_window_configure_request): add hack to
fullscreenize large undecorated windows.
2002-08-12 Havoc Pennington <hp@redhat.com>
* src/stack.c (compute_layer): window is in fullscreen layer if
any member of its group is fullscreen
* src/window.c (meta_window_unmake_fullscreen): update layer for
whole window group
(meta_window_make_fullscreen): ditto
* src/util.c (meta_unsigned_long_hash): move hash/equal funcs for
Window in here.
* src/group.c: track window groups so we can do stuff with them.
2002-08-04 Havoc Pennington <hp@redhat.com>
* src/frames.c (meta_frames_button_press_event): raise/focus
on click, even if the click was on the client area
(this makes Alt+button1 raise windows again, yay)
* src/stack.c (compute_layer): put panels in the DOCK layer always
(keep them on top of other windows). Still sloppy-focus raised
with respect to other docks.
* configure.in: remove -Wshadow for now as GTK headers make all
kinds of noise with it.
2002-07-06 Havoc Pennington <hp@pobox.com>
Apply blackc@speakeasy.net patch, bug #83940, to do
mini-workspaces similar to the pager, when switching
spaces.
* src/window.c (update_net_wm_state): actually fill in
wm_state_skip_taskbar, wm_state_skip_pager flags
* src/tabpopup.c: support drawing a mini-workspace similar to the
one the pager draws.
* src/stack.c (meta_stack_list_windows): new function to list
the windows in stacking order
* src/screen.c (meta_screen_ensure_workspace_popup): don't pass in
the ugly default app icon for workspaces
* src/display.c (event_callback): fix from blackc@speakeasy.net
to avoid dereferencing a NULL grab window.
2002-06-17 Havoc Pennington <hp@pobox.com>
* src/stack.c (meta_stack_get_default_focus_window): don't use a
minimized window as the next focus window, patch from
blackc@speakeasy.net
2002-06-08 Havoc Pennington <hp@pobox.com>
Apply big patch from blackc@speakeasy.net adding a popup window
to the Ctrl+Alt+arrows shortcuts. #83940
2002-06-08 Havoc Pennington <hp@pobox.com>
* src/screen.c (meta_screen_new): select key press/release on the
display->no_focus_window, another attempted fix for not getting
keybindings when no window is focused. Still doesn't seem to work
though. I don't get what's going wrong.
(meta_create_offscreen_window): new function, used instead of
XCreateSimpleWindow so we get override redirect offscreen windows.
2002-05-28 Havoc Pennington <hp@pobox.com>
Patch from Erwann Chenede for raise_or_lower keybinding
* src/display.c, src/common.h: POINT_IN_RECT moved to a common
location, removed from here
(meta_rectangle_intersect): move here and make it public
* src/prefs.c: add raise_or_lower keybinding
* src/stack.c (meta_stack_get_below, meta_stack_get_above): add an
arg to only get windows within the same layer
* src/keybindings.c (handle_raise_or_lower): add handling for a
"raise window if obscured, else lower" keybinding
2002-05-11 Havoc Pennington <hp@pobox.com>
* src/fixedtip.c (meta_fixed_tip_show): keep the tooltip
on the screen horizontally, #76825
* src/window.c (meta_window_handle_mouse_grab_op_event): end grab
op _after_ doing the final update of the move or resize.
Hopefully I didn't have a reason for the order I was using before.
2002-05-10 Havoc Pennington <hp@pobox.com>
* src/tools/metacity-window-demo.c: add override redirect test
window
* src/stack.c (raise_window_relative_to_managed_windows): new
function, used to avoid moving windows above override redirect
popup windows.
* src/display.c (event_callback): don't lower panels on
LeaveNotify if they have focus, #70895
