Applied, thanks!

On Mon, Jan 24 2022 at 15:27:07 +0200, Artur Sinila 
<logarithmus.dev@gmail.com> wrote:
> I've never seen anyone ENCOURAGING me to use goto

In C, goto is and has always been the defacto way of doing error 
handling. You will find this type of error handling everywhere in the 
linux kernel, libwayland, weston, wlroots, sway, gtk, and all of my own 
projects to give just a few examples. The Linux coding guidelines on 
the matter can be seen here: 
https://www.kernel.org/doc/html/v4.17/process/coding-style.html#centralized-exiting-of-functions.

> but will try doing goto & compare line count

Sure, but the goto implementation should be used regardless. :)

> -	return wlrun(config);
> +	return free_and_exit(&config, wlrun(config));

This is a little odd. The conventional way of doing cleanup/error 
handling in C is with gotos, like so:

    int main() {
        int ret = EXIT_FAILURE;
        struct thing *thing = thing_create();
        if (init_thing(thing) == -1) {
            goto error;
        }
        if (setup_thing(thing) == -1) {

Applied, thanks!

(I took the liberty of appending '/' to limit it to folders)

> Useful for laptops with external displays connected. External displays
> usually have an ability to adjust color temperature via built-in menu,
> so you can run `wlsunset` exclusively for laptop's internal display

While external monitors can configure their color temperature, they 
usually do not adjust based on time of day. wlsunset tracks the sun 
after all. The feature is pretty small though, so if it has some use 
that's fine.

Right now the patch is too big, and things like .clang-tidy and use of 
_GNU_SOURCE is a separate topic altogether. I would appreciate if you 
could make a patch containing just the feature and related changes

> Long story short, try setting your launcher as `swaymsg exec --` and 
> it should work fine.

True, that does solve it. The example configs in the repo are still 
just using "swaymsg exec " directly though.

> This patch allows to properly distinguish app path from its 
> arguments, there were a lot of underlying issues which were all 
> caused by this.

Why do you want to distinguish app path from its arguments if the goal 
isn't to call exec(3)?

You would normally pass an unlexed command line to `/bin/sh -c` (just

Hmm, this does appear to cause somewhat annoying behavior. 
Specifically, it breaks use of `swaymsg exec` as a custom launcher, 
which for sway is more or less the suggested custom launcher.

I am considering reverting this patch. What was the use-case that 
required it?

See: https://todo.sr.ht/~kennylevinsen/seatd/10

I'm not sure what I think about ifdef'ing glibc/musl and casting the 
kernel-provided constants around. I suspect we'd be better off just 
disabling Werror on these platforms.

Applied, thanks!

Yeah, there are known issues with Xe and continuous gamma updates. Not 
sure why - I have seen it on a colleagues machine, but not having one 
myself makes it a little hard to debug.

I'm not sure if the issue is wlsunset, wlroots or kernel driver. The Xe 
hardware and kernel driver handles gamma very weird and very 
differently as the hardware implements non-linear gamma LUTs, but the 
user-space API is linear.