On Sun, 19 Mar 2006, Nicholas Cole wrote:
But (back to the original problem) fixing ncurses' port to OS X
would
make this problem moot. I'd like to see if SIGWINCH is defined
or not -
part of seeing where the problem may lie. It is simple to see if
SIGWINCH
was defined when ncurses was compiled since _nc_handle_sigwinch()
would
appear in a "nm" of the ncurses library.
It doesn't appear on an nm of either /usr/lib/libncurses.5.4.dylib or
/usr/lib/libncurses.5.dylib
That sounds like what I was thinking (unless nm isn't returning any
symbols at all- some shared libraries work with nm, some don't).
Most platforms that have ifdef's for POSIX in their headers make
provision for handling the more common non-POSIX extensions. The
way SIGWINCH is ifdef'd in the header I was reading for MacOS X 10.4.3
doesn't leave any way to use _POSIX_C_SOURCE and still use SIGWINCH.
In different versions of the configure script, I've either suppressed
the check for the "darwin" platform, or allowed it to try to set the
standard #define's. Depending on which I choose, one or the other of
the programs using that configure macro works or not.