about keycodes

[chinlu chinawa]

Hello,

I've been trying to find out how does curses do so the
keystroke-getting mechanism bypasses endianness
issues.
 
I hardly tried to setup qemu with some big-endian
linux on it so I could test this, but it's been
completely impossible.

I started then by kgetch, which uses a `struct tries'
which is the one that actually contains each keycode. 

Data for this structure seems to come from the
respective terminal description, which is filled by
newterm through _nc_init_keytries

By tracing a test case, I can get to see this:

char 0 = 033
char 1 = 'O' = 0117
char 2 = 'D' = 0104
char 3 = 000
char 4 = 000
char 5 = 000
char 6 = 000
char 7 = 000
char 8 = 000
char 9 = 000
peeking at 2
ch: 'D' = 0104
ptr=0x808e940, ch=68, value=260
end of sequence
wgetch returning : 'KEY_LEFT' = 0404

I've been trying to make some sense out of this, for
example perhaps values on the fifo were summed, and/or
then (un)mettaed ((x) & (~0x80)), or (un)controlled as
keycodes seem to start over 256 decimal (MIN_KEY 0401)
have tried different things, though I don't seem to
get how it "magically" knows the keypress was
key_left.

I've also been trying to trace the _keytry pointer,
but I don't seem to push the right bits, even when I
can see there is a function which traces it, neither
could access to it by hand via the SCREEN pointer
given by newterm, due to that data being private.

Thanks,




      ___________________________________________________________ 
Yahoo! Mail is the world's favourite email. Don't settle for less, sign up for
your free account today 
http://uk.rd.yahoo.com/evt=44106/*http://uk.docs.yahoo.com/mail/winter07.html