unhandled xterm middle arrow sequence

[anaseto]

Hi,

In xterm, the keypad middle arrow can produce a \e[OE sequence, but it is not parsed by tcell. It seems that there are 4 possible sequences for this key.

[gdamore]

I'm really not sure what to do with this. I'm not even sure how this key should be reported. The lack of consistency is just one issue.

The reality is that many modern keyboards don't even have a separate way to report this. (This is why some keyboards separate cursor keys from the numeric keyboard.)

I'm not trying to be facetious when I say I have no idea what we should even call this keypress. It's not a cursor movement, so what is it?

What terminal emulators actually report meaningful data here?

[anaseto]

With a sequence similar to the arrow key ones, adding t.prepareKey(KeyClear, "\x1bOE") in tscreen.go catches the middle arrow in xterm (another possible sequence is "\x1b[E").

That said, I too cannot find any clear sources on which key should be reported: above I put KeyClear just for testing because that is what I've seen done in a game for the windows console as well as for the SDL interface, but it may hide a hack. The curses library seems to report it consistently as KEY_B2, but it seems that this is a keypad specific code.

[gdamore]

This is kb2 in terminfo. There is also ka1, ka3, kc1, and kc3.

[gdamore]

I'm a little concerned that KeyClear may not accurately describe how users think of this key in all circumstances. Admittedly I don't have a keyboard that has this key.

[anaseto]

Hi, I agree, I don't know how users might think about KeyClear. I must admit that outside terminal games (in particular traditional roguelike games) full access to keypad keys is probably not very useful, so. If something is done, instead of using KeyClear maybe adding a separate constant KeyB2 or Key5 (and the other keypad keys) would be less surprising, but I don't know how it should be reported on systems without terminfo like Windows.

[gdamore]

I'll probably add a specific constant for it.