Signed-off-by: Cameron Nemo <cam@libnemo.org>

Fix #68

Signed-off-by: Cameron Nemo <cam@libnemo.org>

Signed-off-by: Cameron Nemo <cam@libnemo.org>

The word “twilight” is properly translated to ”halvdager”, however
(presumably) as this is an uncommon word in Swedish, the translator
has opted to change it to “dusk” and translate it to ”skymning”,
however in the context it is talking about the twilight in the
morning, so it should have been “dawn”, so the correct translation
would is “gryning”.

Signed-off-by: Mattias Andrée <m@maandree.se>

Signed-off-by: Mattias Andrée <m@maandree.se>

Signed-off-by: Mattias Andrée <m@maandree.se>

Optionally use fnmatch to match output names
Add conffile example for new option

	For big values of `output->gamma_size`, the initialization of
	`ramp_bytes` may cause integer overflow. This commit fixes this
	by switching the type of `size` from `int` to the POSIX type
	`ssize_t` and checking for overflow.

`wayland_set_temperature` assumes that after `wl_display_roundtrip`,
some of the callbacks for `zwlr_gamma_control_v1_listener` will have
been called to initialize `output->gamma_size`.

However, this isn't necessarily true, for example if an output is
plugged in right before `wl_display_roundtrip` is called, that output's
`zwlr_gamma_control_v1_listener` won't have been initialized, thus it
won't get a callback, and when `wl_display_roundtrip` returns,
gammastep will read an uninitialized `gamma_size` (as reported by
Valgrind), and potentially crash if it reads a non-zero value.

I have also seen this uninitialized read happen in other scenarios, for
example when unplugging an output or switching TTYs, sometimes wlroots
will not call any gamma callbacks. But the one about plugging an output
before `wl_display_roundtrip` is consistent (and gammastep's fault).

Initialize `output->gamma_size` during creation to prevent this.

Before this commit, the list of outputs was kept in a dynamic array
that is resized using realloc when an output is added or removed.

This is problematic because pointers to the elements of the dynamic
array are passed to `zwlr_gamma_control_v1_add_listener` so that they
become the `data` argument of the listener's callback. When an output
is added/removed, the dynamic array may get relocated, thus pointers
to other outputs become invalid/dangling, and gammastep may crash.

This can be reproduced quite consistently by adding a `sleep` before
the call to `wl_display_roundtrip` in `wayland_set_temperature`, and
plugging or unplugging an output during the `sleep`.

This commit fixes it by making `outputs` a wl_list, so outputs can be
added or removed without invalidating pointers to other outputs.

Linus Heckemann authored 2 years ago and Cameron Nemo committed 2 years ago

Since commit c5c3ac01 / MR !9, failure
is handled for inputs without any gamma control support such as headless
outputs. However, in some cases, such as when switching VTs, failure is
only temporary.

Implement recovery by resetting output->gamma_control on failure.
On the next call to wayland_set_temperature, we will attempt to reinit
and either recover successfully, or repeat failure.

Fixes #43: Failure after VT switch

Co-authored-by: Joan Bruguera <joanbrugueram@gmail.com>

Fixes #34

When get_gamma_control is invoked on an output which does not support
gamma control (e.g., a headless output), the `gamma_control_failed`
callback gets called (see
<https://github.com/swaywm/wlroots/blob/bfd020047dbe3d3f1a3734d4fe48f04847a46603/types/wlr_gamma_control_v1.c#L214>).

gammastep currently does nothing in this callback, which means that the
gamma_size field of the output stays uninitialized. This causes #3, when
the value in memory happens to be a large number.

This patch sets the gamma_size field to 0 in the failure handler, so
that unsupported outputs are handled correctly.

Thiago Kenji Okada authored 4 years ago and Cameron Nemo committed 4 years ago

The error message popup was showing empty with the following error
message showing in the stderr:

```
</i>' from markup due to error parsing markup: Error on line 2 char 45: “ ” is not a valid character following a “<” character; it may not begin an element name
```

It happening because the error message sometimes contain either < or >
characters, the XML markup was trying to interpret as a tag and failing.

This commit fixes it by escaping the text to be XML safe.

Zsolt Donca authored 4 years ago and Cameron Nemo committed 4 years ago

The affected code checks if there are outputs that support
gamma adjustment, and in case there aren't, it used to log
an error and exit. Since this condition appears to match
when closing the lid, it makes sense to consider this as
just a warning and continue.

Closes #24

libappindicator was deprecated and it will be removed in the
future Debian release.

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=895037

There is a successor API compatible library - Ayatana Indicator.

https://ayatanaindicators.github.io/code/

Predecessor patches/commits:

https://github.com/kenhys/redshift/commit/f2df94b6ad23eceea8c163f9c910e1f6ea122428.patch

https://salsa.debian.org/debian/redshift/-/raw/c70977fea0a55bf14dbbd87ba8ac783ad73c796a/debian/patches/redshift-gtk-use-Ayatana-AppIndicator3-instead-of-App-Ind.patch

The Wayland abstraction is not currently needed, but it may be needed in
the future.

This is puts the data in a strange directory no other Void package uses.
Reverting for now.

This reverts commit 83fd1967.

Thiago Kenji Okada authored 4 years ago and Cameron Nemo committed 4 years ago

This is related to the recent log changes: 6e785e65

The log format changed in two ways:
- Each entry of the log is prefixed by the loglevel
    + *Before*: `Something: Value`
    + *After*: `Loglevel: Something: Value`
- All log output is now in stderr instead of a combination of stdout
  (for info messages) and stderr (for error messages)

However, `gammastep-indicator` needs to parse the log to work, and since
no modification was done in it the output was broken, specially in
"Enable" and "Info" items. This commit fixes it.

Co-authored-by: Cameron Nemo <cnemo@tutanota.com>

closes #21

close #19