x11: Support numpad arrows/Home/End/PageUp/PageDown/Insert/Delete

[forbjok]

Uncommented and corrected x11 key mappings for the numpad keys with num-lock turned off.

[tomaka]

Seems suspicious that this was commented out, but I don't remember why it was the case.

[forbjok]

There are a lot of commented-out x11 keycodes in there. My guess is they were copied from a list at some point, and only the ones deemed of immediate importance were mapped correctly, while the rest were just left commented-out to be fixed later.

That said, after posting this, I found another pull request (#114) that among other things maps some of these, but to other keycodes.

There is some discussion there about whether the numpad keys should map to the same VirtualKeyCode regardless of numlock state. While for my particular case (the terminal alacritty using it for input) this does not matter, I can see this being useful for other purposes, such as games where you might want to use the numpad keys as an additional input separate from the regular arrow keys. In that case, I guess alacritty would have to be changed to use a different method of mapping them to input depending on numlock state.
It really depends on what the intended purpose of VirtualKeyCodes is - to be able to uniquely identify each physical key on the keyboard, or to identify their functions.

However, either way, they should be mapped to something so that it's possible to use them.

EDIT: Also, the old pull request (#114) maps the incorrect codes for PageUp and PageDown. The numpad PageUp and PageDown keys with numlock off actually produce ffi::XK_KP_Prior and ffi::XK_KP_Next, not ffi::XK_KP_Page_Up and ffi::XK_KP_Page_Down.

EDIT2: Nevermind, it seems those have the same underlying int values, so it should make no difference.

[francesca64]

I'm still of the belief that the VirtualKeyCodes should be dependent on numlock state, as numlock changes the semantic meaning of those keys. Someone wishing to use them as directional/etc. input in a game would use scancodes, which refer to physical key positions.

[forbjok]

Yeah, I think I agree on that.

[francesca64]

What's stopping this from being merged? It seems we have a good consensus.

[tomaka]

Needs an entry in the changelog and we're good to go!

[forbjok]

I changed it to use XK_KP_Page_Up and XK_KP_Page_Down instead of Prior and Next, as they have the same underlying code and are better named, and added a changelog entry for the change.

[tomaka]

Thanks!