On Sun, 17 Jan 2010, Ari Pollak wrote:
Thomas Dickey wrote:
...as an example, looking at this chunk in gntkeys.c, almost all of
it is incorrect:
What's incorrect about it, other than not using terminfo?
It makes assumptions about the corresponding terminal descriptions which
don't match any of the ones that I've constructed for ncurses. You might
want to use infocmp on each of those names, and look at the results.
For instance, this line might make sense if "gnt_key_cup" stands for
control-cursor-up:
gnt_key_cup = "\033" "[1;5A";
However, xterm and rxvt don't send the same control sequences for
control-cursor-keys. rxvt and rxvt-unicode send the same, since
almost all of the latter's terminal-behavior (such as this) is from the
former.
screen and rxvt-unicode also differ - screen passes through many of the
control sequences which it doesn't know about. It's possible that the
developer only tested screen with that particular terminal. In this case,
a "screen.rxvt-unicode" entry would be needed to make things work properly
- see the other "screen.XXX" entries that I made to see how it works.
The controls[] and alts[] lists - the latter ignores shift, which is more
common than meta (which still isn't the same as alt).
(the check for null pointer was also redundant, from a quick read of the
code ;-)
For the rest, I'd have to pull out the file again (comments here are just
the obvious stuff that I noticed last week).
--
Thomas E. Dickey
http://invisible-island.net
ftp://invisible-island.net
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
