Tests for sway

[acrisci]

Since sway is officially supported by this library, we should have tests that test it against a running sway binary.

[nwg-piotr]

On Arch Linux I use an old -git package, which includes the "add event_state_mask and symbols to binding info" commit, and it works well. Tonight I built a Void Linux package v2.0.1, and it crashes for me on the 'binding' event again. For now no idea what's happening. On i3 everything works well.

[acrisci]

The binding event works fine for me.

[nwg-piotr]

Strange... I'll try to take a closer look on Arch tonight.

[nwg-piotr]

Alright, there was a mess in virtual environments. After cleaning it up, I'm having some other difficulties with my code, but the binding event does not crash any longer. Sorry for my mistake.

[Galicarnax]

When running under i3, i3ipc.Connection().get_tree() returns a single object representing the root container. When used under sway, get_tree() returns a linear list of all containers (root, workspaces, windows). Using i3ipc version 2.2.1-1 under Arch in both cases. No mention in the docs that get_tree() might behave differently under i3 and sway.

[Galicarnax]

Never mind, it was an issue with a 3rd-party library, not i3ipc-python. Sorry.