hl.dsp.window.fullscreen_state is broken

[lerobynetcool]

I observe a regression from 0.54 to 0.55

The following should put the window in tiled mode, while keeping the client's fullscreen state, but this is no longer respected.

hl.dsp.window.fullscreen_state({
  client=-1,
  internal=0,
  action="set"
})

The issue is that client=-1 behaves as if it were client=0

A workaround is something like this

hl.dsp.window.fullscreen_state({
  client=hl.get_active_window().fullscreen_client,
  internal=0,
  action="set"
})

As discussed on the discord channel, maybe the API of fullscreen_state could be improved by making both internal and client optional : they could behave by default as -1

[Mervius]

It seems the internal argument in the dispatcher is also broken.

hl.bind(mainMod .. " + SHIFT + F", hl.dsp.window.fullscreen_state({ internal=hl.get_active_window().fullscreen,client=hl.get_active_window().fullscreen_client,action = "set" }))

This bind should leave the fullscreenstate unchanged but it fullscreens any non-fullscreen window

[Mervius]

However, using -1 in the internal state instead of hl.get_active_window().fullscreen then un-fullscreens fullscreen windows and does nothing to non-fullscreen windows...

Even though those two should return the same value

[lerobynetcool]

Here is the full workaround I'm currently using. I simplified it to a one liner in my post, but didn't test it.

local function hypr_full(mode,float)
  return function()
    local c = hl.get_active_window().fullscreen_client
    hl.dispatch(hl.dsp.window.fullscreen_state({client=c,internal=mode,action="set"}))
    hl.dispatch(hl.dsp.window.float({action=float}))
  end
end

hl.bind("SUPER+ALT     +F", hypr_full(0,"on"))
hl.bind("SUPER+CONTROL +F", hypr_full(2,"off"))
hl.bind("SUPER         +F", hypr_full(1,"off"))
hl.bind("SUPER+SHIFT   +F", hypr_full(0,"off"))

[LRitzdorf]

I encountered a version of this as well, wheremy "toggle client fullscreen" bind turned into a "set client fullscreen" one. Solved it by using:

hl.dsp.window.fullscreen_state({ action = "toggle", internal = 0, client = 2 })

in my bind. This is indeed different from my original setup under Hyprlang, which used -1 instead of 0.

Something to consider: having a "toggle" action and a -1 "keep previous value" flag is a little weird. It feels like there should be a better way to format this, but I'm not sure precisely what that'd be.

[dstrelnikov]

Hey, thanks for your fix! I encountered the same problem. To reproduce the logic of:

bind = Super Shift, F, fullscreenstate, -1 2

I have to use

hl.bind( "SUPER + SHIFT + F", hl.dsp.window.fullscreen_state( { internal = 0, client = 2, action = "toggle" } ))

It also won't work for me without action="toggle" which does not align with the docs:
action toggle(default if no value given), enable/on, disable/off

Correction: it doesn't reproduce the same logic but somewhat works for me.

[kolayne]

Hey! While we are at fixing the fullscreen state bugs, I also suggest to rethink the fullscreen state logic a bit. Please, take a look at #14646 and share your thoughts: whether the proposed behavior makes sense or if you'd expect something else.