1 - A Key corresponds to a single keypress by the user:
2 - A single ASCII byte, such as "a" (97)
3 - A series of bytes corresponding to a single Unicode character, such
4 as "æ" (195,166), "Ꮬ" (225,143,156), or "" (240,144,144,166)
5 - A Sequence is a specific series of Keys
6 - An escape sequence in an editor, such as:
8 - A cheat sequence in a game, such as:
9 →↑→ (M[C-M[A-M[C) ((27,91,67), (27,91,65), (27,91,67))
11 - Not sure how to handle the fact that Unicode needs multiple bytes of input
12 - Will I have to have some sort of global loop?
13 - Time pauses 'tween bytes to form Key objects, synchronize `doupdate` calls
16 - Ruby Threads? Seriously? >,>
17 - What about a *single*, *central* global loop?
18 - Something like Rack that handles "the Loop" for everything that needs it
19 (EventMachine, XMPP4R, Rack, and Nfoiled?)
20 - Any library can schedule things to work during the loop, all must be non-
22 - What about popping a thread everytime a non-ASCII byte is read?
23 - Do all Unicode characters start with a non-ASCII byte? How are they
24 sequential? (for #succ, etc)
25 - I could pop a `Thread`, and instruct it to `gets` until it receives a
26 nil, indicating the byte sequence is over.
27 - Do Unicode characters of different byte-lengths pad with nils to a
28 specific number of bytes? Try turning back on the display of `nil`s
30 - Do all terminals send the bytes of a unicode character fast enough to
31 prevent intervening cyclings of `gets`? Must experiment on a really
33 - What if you were to provide input so fast as to prevent an successive
34 nil? Is that possible? Must experiement with pasting Unicode
35 characters/typing very fast.
36 - If for some reason the user had turned back on `delay` mode or set a
37 non-zero`wtimeout`, we'd have to retreive and save that value, then
38 switch into `nodelay` mode; retreive the Unicode bytes; and re-set
39 their preferred input mode.
40 - If we're already in a `gets` loop ... we'd have to have an 'input
41 lock' of some sort, to ensure that the `gets` from the original input
42 loop doesn't accidentally 'steal' a byte from the middle of the
43 Unicode series. (and recognize it as a Unicode byte, and start a new
44 `Thread`, and ... so on)
45 - Speed concerns involving popping a new thread for every Unicode
46 character. Will have to benchmark a bit.
47 - Would mean we couldn't handle `doupdate` calls in a sexy way ... have to
48 leave that up to the user. Which is probably fine anyway, I guess, just