ghgemmer@msn.com wrote: > Since I'm using LynxOS 4.0 are there any manual modifications to > create a configuration file that I need to do. I read the INSTALL and > TODO files but didn't see anything I needed to do. Are there some > checks that the configure does that don't work automatically for > LynxOS 4.0 system that I need to manually fiddle with. > cd ../objects; gcc -I../progs -I. -DNDEBUG -I. -I../include -I/usr/ > local/ncurs > es/include/ncurses -DHAVE_CONFIG_H -O2 -c ../progs/tset.c > ../progs/tset.c: In function `get_termcap_entry': > ../progs/tset.c:525: warning: assignment makes pointer from integer > without a cast > ../progs/tset.c:526: dereferencing pointer to incomplete type > ../progs/tset.c: In function `reset_mode': > ../progs/tset.c:726: `BSDLY' undeclared (first use in this function) > ../progs/tset.c:726: (Each undeclared identifier is reported only once > ../progs/tset.c:726: for each function it appears in.) > ../progs/tset.c:726: `VTDLY' undeclared (first use in this function) > ../progs/tset.c:728: `FFDLY' undeclared (first use in this function) Judging by the headers on Solaris, they're BSD-symbols rather than POSIX, and probably should be ifdef'd just in case. Right now, they're associated with another symbol: #ifdef NLDLY | NLDLY | CRDLY | TABDLY | BSDLY | VTDLY | FFDLY #endif -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net

ghgemmer@msn.com wrote: >> - Show quoted text - > Thomas > Thanks for the quick response. I didn't expect to get a response from > a main ncurses contributor. google shows me the comment (though google's been not very well the past couple of months - it omits most of the interesting articles). So I added this group to my newsreader, and followed-up. > I'll doing what you indicated above. > Even with the above it was able to build a libncurses.a file which I > thought would be good enough to try some things out. However when I > compiled and linked the following simple program and then ran it > with arguements e 10 10 10 10 I got the following output which is > not what the program is intended to produce which is a box of width 10 > and height 10 at position 10,10 on the screen. > eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee I see ("e" is supposed to be repeated, and the cursor movement is not happening). > One thing that may/may not make a difference is that I was using gcc > 2.95.3 to make the ncurses library on the target but using Windows XP > with cygwin and gcc 3.2.2 to compile the test program below and link > on the host using the generated ncurses library transferred from the > target . gcc 2.95.3 is what I currently have on the target system and > gcc 3.2.2 is the cross compiler I'm using on the windows system. I > was going to try to either fully compile in one or the other as my > next attempt. That might be more reliable. I'm not aware of any previous port of ncurses to LynxOS, but if it compiles without a lot of problems, that's promising. Generally I configure using "--with-trace" and turn on the trace feature in the library to see what it thinks it's doing. Your one-liner could be some obscure data issue (can't think what offhand). > Also if I just use the curses library supplied with LynxOS (this is > pre ncurses and doesn't have color support or alot of other things) it > wil draw the box correctly. -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net

ghgemmer@msn.com wrote: > I modified two files to use conditional compilation on those symbols > and then everything except the C++ stuff compiled ok, including the > test programs. I don't need the C++ stuff at this point. The > compilation was done using gcc 2.95.3. I suspect the C++ problems were with the header files (which are not always in sync with the compiler). For the compiler itself, I have built with that version of g++ (some warnings, but workable). > I then ran some of the test programs and they pretty much output > gibberish on the screen. > I then started fiddling with the terminal type thinking that may be > the problem and finally got most of the test programs to work with one > of the terminal types by setting TERM to that terminal type name. So > I think it is just a matter looking at the /etc/termcap file to make > sure all the appropriate attributes are specified correctly for the > terminal I am using. > Apparently none of the terminal defs in the default /etc/termcap > specify any terminal as having colors so I'm in the process of adding > in a terminal definition for the TN3270 terminal emulator I am > telneting in from. That's a rather unusual choice (why not a vt100?) -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net