Is it true that pthreads on linux uses pipes for communication between main
thread and other threads ?
However, it is true that Linux pthreads implementation uses a pipe
for communication between all "user" threads (including the initial
thread) and the (automagically created) manager thread. The latter
is responsible (among other things) for thread creation, signal
propagation, process termination, etc.
In order to understand recursion you must first understand recursion.
: I've got some annoying bug with my application. ALL windows I open have a
: keypad(win, TRUE) command to allow cursor keys. Sometimes, if I type the up
: cursor arrow while the screen is refreshed, I get [ESC]A in the window buffer
: as TWO separate characters instead of one integer (KEY_UP). As [ESC] is define
Curses has a little timing trouble in the ESC sequence detection area.
One is that it does a sleep(1) to see if the next character coming in is
part of the sequence... sleep(1) might be as little as 1/100 of a second.
The other is that it doesn't properly block if you have another flag set,
the name of which escapes me at the moment, I think nodelay(...,TRUE).
This is usually only a problem across some type of buffered interface, like
TCP, or a compressing modem, possibly a smart serial card with buffers.
SCO had a patch, which really only consisted of lengthening the sleep to 8
seconds, but then your legitimate ESC would take 8 seconds to have an
Unisys had a replacement libcurses, and I think that SVR4.2 should have the
problem resolved by some different method.
- Pope Valley & Napa CA.