[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: LYNX-DEV 2.8pre.4
From: |
Jonathan Sergent |
Subject: |
Re: LYNX-DEV 2.8pre.4 |
Date: |
Fri, 6 Mar 1998 20:04:38 -0500 |
Dave Eaton <address@hidden> wrote:
>On Fri, 6 Mar 1998, T.E.Dickey wrote:
>
>> This fixes the bugs reported in the pre-release of Lynx 2.8 (are there any
>> more?)
>
>Woo hoo, built and ran on HP-UX 10.20 (PA RISK 1.1) *without* needing
>ncurses! Thanks!
Spiffy -- I wanted to make sure that the configure changes I suggested
were appropriate for HP-UX 10.20 in general and not just our local setup
and weren't going to break things further for other people.
The patch checks for hpux10.* -- I'm curious as to whether this works
on other 10.x's or if they have other oddities.
I want to mention that I get this weirdness using gcc on HP-UX in
config.log with -I/usr/include/curses_colr:
configure:6869: checking if curses supports color attributes
configure:6890: gcc -o conftest -I/usr/include/curses_colr -O2 -DSNAKE
conftest.c -lcur_colr 1>&5
In file included from configure:6877:
/usr/include/curses_colr/curses.h:325: parse error before `va_list'
/usr/include/curses_colr/curses.h:326: parse error before `va_list'
configure: failed program was:
#line 6875 "configure"
#include "confdefs.h"
#include <curses.h>
int main() {
chtype x = COLOR_BLUE;
has_colors();
start_color();
#ifndef NCURSES_BROKEN
wbkgd(curscr, getbkgd(stdscr)); /* X/Open XPG4 aka SVr4 Curses */
#endif
; return 0; }
This happens with many of the configure tests.
This is quite possibly ejbecause the gcc installation on these machines
may be severely messed up. It works fine with the HP [unbundled]
C compiler, which I had been using before for other reasons.
Perhaps we just need someone to run fixincludes correctly here after
they upgraded the OS. .../gcc-lib/.../include/ has a curses.h but not
a curses_colr/curses.h, which leads me to believe that's to blame.
Perhaps confdefs.h just needs a #include <varargs.h>.
The gcc generated binary works, just no color because the config tests
all fail for the wrong reason (above parse errors which look (from
inspecting the header files) to be related to varargs somehow.
I think it's something local since you got it to work with gcc out of
the box on your end. Or did you not get color either? Check your
config.log and see if (a) it is using -lcur_colr, and (b) if it thinks
it has color support.
--
Jonathan Sergent / address@hidden