- drw: minor improvement to the nomatches cache
- overhaul utf8decoding and render invalid utf8 sequences as U+FFFD.
Thanks NRK for these improvements!
The Makefile used to suppress output (by using @), so this target made sense at
the time.
But the Makefile should be simple and make debugging with less abstractions or
fancy printing. The Makefile was made verbose and doesn't hide the build
output, so remove this target.
Prompted by a question on the mailing list about the options target.
From sigaction(2):
A child created via fork(2) inherits a copy of its parent's signal dispositions.
During an execve(2), the dispositions of handled signals are reset to the default;
the dispositions of ignored signals are left unchanged.
This refused to start directly some programs from configuring in config.h:
static Key keys[] = {
MODKEY, XK_o, spawn, {.v = cmd } },
};
Some reported programs that didn't start were: mpv, anki, dmenu_extended.
Reported by pfx.
Initial patch suggestion by Storkman.
SA_NOCLDWAIT is marked as XSI in the posix spec [0] and FreeBSD and NetBSD
seems to more be strict about the feature test macro [1].
so update the macro to use _XOPEN_SOURCE=700L instead, which is equivalent to
_POSIX_C_SOURCE=200809L except that it also unlocks the X/Open System
Interfaces.
[0]: https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/signal.h.html#tag_13_42
[1]: https://lists.suckless.org/dev/2302/35111.html
Tested on:
* NetBSD 9.3 (fixed).
* FreeBSD 13 (fixed).
* Void Linux musl.
* Void Linux glibc.
* OpenBSD 7.2 (stable).
* Slackware 11.
Reported-by: beastie <pufferfish@riseup.net>
signal() semantics are pretty unclearly specified. For example, depending on OS
kernel and libc, the handler may be returned to SIG_DFL (hence the inner call
to read the signal handler). Moving to sigaction() means the behaviour is
consistently defined.
Using SA_NOCLDWAIT also allows us to avoid calling the non-reentrant function
die() in the handler.
Some addditional notes for archival purposes:
* NRK pointed out errno of waitpid could also theoretically get clobbered.
* The original patch was iterated on and modified by NRK and Hiltjo:
* SIG_DFL was changed to SIG_IGN, this is required, atleast on older systems
such as tested on Slackware 11.
* signals are not blocked using sigprocmask, because in theory it would
briefly for example also ignore a SIGTERM signal. It is OK if waitpid() is (in
theory interrupted).
POSIX reference:
"Consequences of Process Termination":
https://pubs.opengroup.org/onlinepubs/9699919799/functions/_Exit.html#tag_16_01_03_01
It's not uncommon for one keysym to map to multiple keycodes. For
example, the "play" button on my keyboard sends keycode 172, but my
bluetooth headphones send keycode 208, both of which map back to
XF86AudioPlay:
% xmodmap -pke | grep XF86AudioPlay
keycode 172 = XF86AudioPlay XF86AudioPause XF86AudioPlay XF86AudioPause
keycode 208 = XF86AudioPlay NoSymbol XF86AudioPlay
keycode 215 = XF86AudioPlay NoSymbol XF86AudioPlay
This is a problem because the current code only grabs a single one of
these keycodes, which means that events for any other keycode also
mapping to the bound keysym will not be handled by dwm. In my case, this
means that binding XF86AudioPlay does the right thing and correctly
handles my keyboard's keys, but does nothing on my headphones. I'm not
the only person affected by this, there are other reports[0].
In order to fix this, we look at the mappings between keycodes and
keysyms at grabkeys() time and pick out all matching keycodes rather
than just the first one. The keypress() side of this doesn't need any
changes because the keycode gets converted back to a canonical keysym
before any action is taken.
0: https://github.com/cdown/dwm/issues/11
This reverts commit c2b748e793.
Revert back this change. It seems to not be an edge-case anymore since
multiple users have asked about this new behaviour now.
Reasoning: Since 2011 dmenu has been capable of working out which
monitor currently has focus in a Xinerama setup, making the use
of the -m flag more or less redundant.
This is easily demonstrated by using dmenu in any other window
manager.
There used to be a nodmenu patch that provided these changes:
https://git.suckless.org/sites/commit/ed68e3629de4ef2ca2d3f8893a79fb570b4c0cbc.html
but this was removed on the basis that it was very easy to work
out and apply manually if needed.
The proposal here is to remove this dependency from dwm. The
mechanism of the dmenumon variable could be provided via a patch
if need be.
The edge case scenario that dmenu does not handle on its own, and
the effect of removing this mechanism, is that if the user trigger
focusmon via keybindings to change focus to another monitor that
has no clients, then dmenu will open on the monitor containing the
window with input focus (or the monitor with the mouse cursor if
no windows have input focus).
If this edge case is important to cover then this can be addressed
by setting input focus to selmon->barwin in the focus function if
there is no client to give focus to (rather than giving focus back
to the root window).
The purpose and reasoning behind the bar layout width (blw) variable
in dwm the way it is today may not be immediately obvious.
The use of the variable makes more sense when looking at commit
2ce37bc from 2009 where blw was initialised in the setup function
and it represented the maximum of all available layout symbols.
for(blw = i = 0; LENGTH(layouts) > 1 && i < LENGTH(layouts); i++) {
w = TEXTW(layouts[i].symbol);
blw = MAX(blw, w);
}
As such the layout symbol back then was fixed in size and both drawbar
and buttonpress depended on this variable.
The the way the blw variable is set today in drawbar means that it
merely caches the size of the layout symbol for the last bar drawn.
While unlikely to happen in practice it is possible that the last bar
drawn is not that of the currently selected monitor, which can result
in misaligned button clicks if there is a difference in layout symbol
width between monitors.
This is a follow-up on this thread:
https://lists.suckless.org/hackers/2208/18462.html
The orginal code had constraints such that if a window's starting
attributes (position and size) were to place the window outside of
the edges of the monitor, then the window would be moved into view
at the closest monitor edge.
There was an exception to this where if a top bar is used then the
window should not obscure the bar if present, which meant to place
the window within the window area instead.
The proposed change here makes it the general rule that floating
windows should spawn within the window area rather than within the
monitor area. This makes it simple and consistent with no
exceptions and it makes the intention of the code clear.
This has the benefit of making the behaviour consistent regardless
of whether the user is using a top bar or a bottom bar.
Additionally this will have an effect on patches that modify the
size of the window area. For example if the insets patch is used to
reserve space on the left hand side of the monitor for a dock or a
vertical bar then new floating clients will not obscure that area.
The reasoning behind the original line may be lost to time as
it does not make much sense checking the position on the x-axis
to determine how to position the client on the y-axis.
In the context of multi-monitor setups the monitor y position
(m->my) may be greater than 0 (say 500), in which case the window
could be placed out of view if:
- the window attributes have a 0 value for the y position and
- we end up using the y position of bh (e.g. 22)
If the aim is to avoid a new floating client covering the bar then
restricting y position to be at least that of the window area
(m->wy) should cover the two cases of using a top bar and using a
bottom bar.
main change here is making the `zoom()` logic saner. the rest of the
changes are just small stuff which accumulated on my local branch.
pop() must not be called with NULL. and `zoom()` achieves this, but in a
very (unnecessarily) complicated way:
if c == NULL then nexttiled() will return NULL as well, so we enter this
branch:
if (c == nexttiled(selmon->clients))
in here the !c check fails and the function returns before calling pop()
if (!c || !(c = nexttiled(c->next)))
return;
however, none of this was needed. we can simply return early if c was NULL.
Also `c` is set to `selmon->sel` so we can use `c` in the first check
instead which makes things shorter.
when calling die and the last character of the string corresponds to
':', die() will call perror(). See util.c
Also change EXIT_SUCCESS to EXIT_FAILURE
This is in particular to avoid flickering in dwm (and high CPU usage)
when hovering the mouse over a tabbed window that was previously
managed by dwm.
Consider the following two scenarios:
1)
We start tabbed (window 0xc000003), tabbed is managed by the
window manager.
We start st being embedded into tabbed.
$ st -w 0xc000003
What happens here is that:
- tabbed gets a MapRequest for the st window
- tabbed reparents the st window
- tabbed will receive X events for the window
The window manager will have no awareness of the st window and the
X server will not send X events to the window manager relating to
the st window.
There is no flickering or any other issues relating to focus.
2)
We start tabbed (window 0xc000003), tabbed is managed by the
window manager.
We start st as normal (window 0xd400005).
What happens here is that:
- the window manager gets a MapRequest for the st window
- dwm manages the st window as a normal client
- dwm will receive X events for the window
Now we use xdotool to trigger a reparenting of the st window into
tabbed.
$ xdotool windowreparent 0xd400005 0xc000003
What happens here is that:
- tabbed gets a MapRequest for the st window
- tabbed reparents the st window
- the window manager gets an UnmapNotify
- the window manager no longer manages the st window
- both the window manager and tabbed will receive X events
for the st window
In dwm move the mouse cursor over the tabbed window.
What happens now is that:
- dwm will receive a FocusIn event for the tabbed window
- dwm will set input focus for the tabbed window
- tabbed will receive a FocusIn event for the main window
- tabbed will give focus to the window on the currently selected
tab
- which again triggers a FocusIn event which dwm receives
- dwm determines that the window that the FocusIn event is for
(0xd400005) is not the currently selected client (tabbed)
- dwm sets input focus for the tabbed window
- this causes an infinite loop as long as the mouse cursor hovers
the tabbed window, resulting in flickering and high CPU usage
The fix here is to tell the X server that we are no longer interested
in receiving events for this window when the window manager stops
managing the window.
Commit 8806b6e237 ("manage: propertynotify: Reduce cost of unused size
hints") mistakenly removed an early size hints update that's needed to
populate c->isfixed for floating checks at manage() time. This resulted
in fixed (size hint min dimensions == max dimensions) subset of windows
not floating when they should.
See https://lists.suckless.org/dev/2204/34730.html for discussion.
When monitors are removed, the coordinates of existing monitors may
change, if the removed monitors had smaller coordinates than the
remaining ones.
Remove special case handling so that the same update-if-necessary loop
is run also in the case when monitors are removed.
This patch defers all size hint calculations until they are actually
needed, drastically reducing the number of calls to updatesizehints(),
which can be expensive when called repeatedly (as it currently is during
resizes).
In my unscientific testing this reduces calls to updatesizehints() by
over 90% during a typical work session. There are no functional changes
for users other than an increase in responsiveness after resizes and
a reduction in CPU time.
In slower environments or X servers, this patch also offers an
improvement in responsiveness that is often tangible after resizing a
client that changes hints during resizes.
There are two main motivations to defer this work to the time of hint
application:
1. Some clients, especially terminals using incremental size hints,
resend XA_WM_NORMAL_HINTS events on resize to avoid fighting with the
WM or mouse resizing. For example, some terminals like urxvt clear
PBaseSize and PResizeInc during XResizeWindow and restore them
afterwards.
For this reason, after the resize is concluded, we typically receive
a backlogged XA_WM_NORMAL_HINTS message for each update period with
movement, which is useless. In some cases one may get hundreds or
thousands of XA_WM_NORMAL_HINTS messages on large resizes, and
currently all of these result in a separate updatesizehints() call,
of which all but the final one are immediately outdated.
(We can't just blindly discard these messages during resizes like we
do for EnterNotify, because some of them might actually be for other
windows, and may not be XA_WM_NORMAL_HINTS events.)
2. For users which use resizehints=0 most of these updates are unused
anyway -- in the normal case where the client is not floating these
values won't be used, so there's no need to calculate them up front.
A synthetic test using the mouse to resize a floating terminal window
from roughly 256x256 to 1024x1024 and back again shows that the number
of calls to updatesizehints() goes from over 500 before this patch (one
for each update interval with movement) to 2 after this patch (one for
each hint application), with no change in user visible behaviour.
This also reduces the delay before dwm is ready to process new events
again after a large resize on such a client, as it avoids the thundering
herd of updatesizehints() calls when hundreds of backlogged
XA_WM_NORMAL_HINTS messages appear at once after a resize is finished.
In certain instances trans may be set to a window that doesn't actually
map to a client via wintoclient; in this case it doesn't make sense
to set isfloating/oldstate since trans is essentially invalid in that
case / correlates to the above condition check where trans is set /
XGetTransientForHint is called.
maybe leak isn't the best word, given that the object lives for the
entire duration of the program's lifetime.
however, all elements of scheme are free-ed, can't think of any reason
why scheme itself should be an exception.
I noticed that a non-trivial amount of dwm's work on my machine was from
drw_text, which seemed weird, because I have the bar disabled and we
only use drw_text as part of bar drawing.
Looking more closely, I realised that while we use m->showbar when
updating the monitor bar margins, but don't skip actually drawing the
bar if it is hidden. This patch skips drawing it entirely if that is the
case.
On my machine, this takes 10% of dwm's on-CPU time, primarily from
restack() and focus().
When the bar is toggled on again, the X server will generate an Expose
event, and we'll redraw the bar as normal as part of expose().
This reverts commit 716233534b.
It causes issues with truncation of characters when the text does not fit and
so on. The patch should be reworked and properly tested.
Calculates len & ew in drw_font_getexts loop by incrementing instead of
decrementing; as such avoids proportional increase in time spent in loop
based on provided strings size.
It generally doesn't make much sense to allow focusstack() to navigate
away from the selected fullscreen client, as you can't even see which
client you're selecting behind it.
I have had this up for a while on the wiki as a separate patch[0], but
it seems reasonable to avoid this behaviour in dwm mainline, since I'm
struggling to think of any reason to navigate away from a fullscreen
client other than a mistake.
0: https://dwm.suckless.org/patches/alwaysfullscreen/
Many users new to dwm find themselves caught out by being kicked out to the login manager (dwm crashing) when they open 50+ clients for demonstration purposes. The number of clients reported varies depending on the resolution of the monitor.
The cause of this is due to how the default tile layout calculates the height of the next client based on the position of the previous client. Because clients have a minimum size the (ty) position can exceed that of the window height, resulting in (m->wh - ty) becoming negative. The negative height stored as an unsigned int results in a very large height ultimately resulting in dwm crashing.
This patch adds safeguards to prevent the ty and my positions from exceeding that of the window height.
This jarred me a bit while reading the code, since "sw" usually refers
to the global screen geometry, but in drawbar() only it refers to
text-related geometry. Renaming it makes it more obvious that these are
not related.
There are two places that mfact can be set:
- In the mfact global, which is defined at compile time and passed
into m->mfact during monitor setup. No bounds checks are performed,
but the comment alongside it says that valid values are [0.05..0.95]:
static const float mfact = 0.55; /* factor of master area size [0.05..0.95] */
- By setmfact, which adjusts m->mfact at runtime. It also does some
minimum and maximum bounds checks, allowing [0.1..0.9]. Values outside
of that range are ignored, and mfact is not adjusted.
These different thresholds mean that one cannot setmfact 0.95 or 0.05,
despite the comment above that lists the legal range for mfact.
Clarify this by enforcing the same bounds in setmfact at runtime as
those listed for mfact at compile time.
Hiltjo Posthuma
Pontus Stenetorp
NRK
Chris Down
Stein
explosion-mental
Santtu Lakkala
Miles Alan
Quentin Rameau
Ian Remmler
Alex Flierl
Anselm R Garbe
Klemens Nanni