Curses App and Cproto

hi
is there any way to port unix applications which depend on libCurses ?
i’d like, for example, to have nano on RiscOS. Nano depends on libCurses.

also, is there a port of dev-util/cproto for RiscOS ?

sys-libs/ncurses
Homepage: http://www.gnu.org/software/ncurses/ http://dickey.his.com/ncurses/
Description: console display library
License: MIT

app-editors/nano
Homepage: http://www.nano-editor.org/ http://www.gentoo.org/doc/en/nano-basics-guide.xml
Description: GNU GPL’d Pico clone with more functionality
License: GPL-3

dev-util/cproto
Homepage: http://invisible-island.net/cproto/
Description: Generate C function prototypes from C source code
License: public-domain

is there any way to port unix applications which depend on libCurses ?

The problem is Linux has a proper terminal to run these things in. The only terminal we have is in !Nettle (Ansi Task) No support for UTF-8, though it looks like nano can option that off.

I have launched !Nettle, did ssh connect to a linux machine (i386), and then launched nano: the ENTER key is not working at all with nano.

I’m intrigued why you would want to run nano (this isn’t a vi/emacs/pico debate, there’s no point, vi(m) wins) on RISC OS rather than !Zap (Obviously you wouldn’t run StrongEd ;-) )
I am a big, big fan of consoles, I even wrote my own editor on SCO UNIX (laugh all you want) many years ago.

As for the enter problem, have you checked your TERM setting is correct (i.e. the same on nettle as at the command line)
I’d suggest start with vt100 and work your way up to ansi/linux/xterm.
Unless someone else has come across this before.

I prefer nano as terminal editor, it’s the gentoo default, also i want to develop my Apps using Curses

I have implemented an editor from scratches, it is currently also used as “code flow” in a debug processor’s TAP.

btw, i have already tried { vt100, vt220, linux } in Nettle without any progress: nano is still not working on the remote linux machine

is there any port of Cproto ?

i am reading this discussion about all troubles about input

My favourite UNIX text editor is e3—a small binary which can be called by several different names which give different keystroke commands. The default is WordStar; e3me gives Emacs key bindings, while I would usually use e3ne which uses NEdit key bindings: Home goes to start of line, End to end of line, Ctrl-s saves and so on (PC standard but not RISC OS).

But this is one editor which cannot be ported to RISC OS. It’s written in X86 assembler.

Ah! termios.h, that brings back scary memories!
I had no idea that Nettle could do a task window!
It would be nice if it worked as well as a console though!
That’s a nice little editor!

I can’t really drive PICO/nano, not done for ages (20 years).
What do you mean exactly it doesn’t work?
I can type a line press enter and type another line below it.
The shell, htop, iotop and vi works ok and iftop seems to work OK – using vt220.
elinks, kind of works….
I tested connecting to an intel box at home running linux mint.

nedit, the daddy of editors ;-)

btw, i have already tried { vt100, vt220, linux } in Nettle without any progress: nano is still not working on the remote linux machine

I haven’t a SSH server running at the moment. You could try Ctrl j or Ctrl m instead of enter?
Nettle doesn’t have many user controls on its own, like you would get in a serial console. The line editor option allows more intelligent line editing, it is a little strange after using terminals that just work. I think that you might have to configure a mismatch for keys on the nano machine if you can’t find a way that sends the enter key. It’s not the one that gives a problem usually. Backspace/Delete/Left/Right/Up/Down are the usual problems.
I wonder how far nano would get without the esc sequences?
I use a standard taskwindow to SSH and then by issuing sh as a command, causing the remote machine to use a raw shell. (Debian/Ubuntu derivatives, possibly others)

Edit: a further thought, the remote nano may be relying on/expecting UTF-8?

I havent’ tried yet because I am far from my RiscPC (due to my job I need to drive a lot), it may be it’s UTF-8, it may be it’s an other trouble. I will try as soon as i will be back home.

btw, is there any port of dev-util/cproto ? I really need it.

btw, is there any port of dev-util/cproto ? I really need it.

I’ve never used it but cproto crosscompiled in a blink. I think though, it would be a good one for you to compile in RISC OS so you can make changes as you see fit. There is a pipe implentation using tempfile for the DOS port that could be useful in RISC OS for example.
The easy way to do this is to have the cross-compiler autoconfig the source then change the $CC references to ~/gccsdk/env/gcc to plain gcc (etc) in the Makefile.
Do you have an email, as there are changes to allow Flex and Bison to work that aren’t downloadable yet, and I can send you what is needed for that also.
Later, if you get it running to your satisfaction, patches can be put in the autobuilder for ongoing development/downloading.
It is ideal for someone like yourself who is familiar with the program to establish faults/fixes, but help is at hand if needed.

a further thought, the remote nano may be relying on/expecting UTF-8?

Very unlikely, I’d have thought, that this could be cause of the trouble. UTF-8 and ASCII are identically coded in the ASCII range of characters 0 to 127. It would only be an issue if you try typing pound signs (£) or anything else outside 7-bit ASCII range.

Very unlikely, I’d have thought, that this could be cause of the trouble.

Thanks for the input, I can only add that the linux terminal may have some settings, but by the time nano runs it probably requires a particular environment and it could change.

@Ronald May
my email is “carlojpisani at gmail dot com”
thank you =)

i have prepared a pretty launcher, which defines alias with the proper wimpslot of 8Myte, but cproto calls cpp (i have defined an other alias for it, cpp is contained in !gcc) and then it crashes with unexpected behavior.

i will post the error-sentence it provides

Edit: Mmmm -might have revealed an issue with 4.7.4 here, 4.1.2 cpp/cc1 behaves very well with a huge number of c files and a modest wimpslot of 8000k
Seems like 4.7.4 cpp/cc1 is not recovering memory (from it’s already large wimpslot of 15000K or larger) after each pipe instance or something.

hi
i do not get any email

I’ve sent another brief one, see if you get this one.
I found cproto gave a few problems with gcc 4.7.4 but is ok with 4.1.2, and it should be OK with gcc 3 if thats what you need to use.
When cpp is not used, cproto gets confused about <stdio.h>type lines for some reason, otherwise there is no reason to use 2>/dev/null to hide the errors.
Edit: there is the -q option to suppress those errors. It only happens when cpp isn’t used (and the directorys aren’t included)

i see your email, and you should have already got a reply

typing “cproto hAllo.c” it claims

UNIXLIB detected recursion of signal SIGSEGV exiting
UNIXLIB detected recursion of signal SIGSEGV exiting

(repeated twice)

and it stops

Googling for SIGSEGV I found

This signal is generated when a program tries to read or write outside the memory that is allocated for it, or to write memory that can only be read. (Actually, the signals only occur when the program goes far enough outside to be detected by the system’s memory protection mechanism.). The name is an abbreviation for “segmentation violation”. Common ways of getting a SIGSEGV condition include dereferencing a null or uninitialized pointer, or when you use a pointer to step through an array, but fail to check for the end of the array. It varies among systems whether dereferencing a null pointer generates SIGSEGV or SIGBUS.

that sounds like, ah, the segmentation fault, the bane of every beginning C programmer’s existence.

I am reading about UnixLib wondering what is wrong with my setup.

UNIXLIB detected recursion of signal SIGSEGV exiting

I think you can get this when running out of memory (wimpslot). How much total/free memory do you have? Can you run gcc 4.1.2? Some ported programs crash immediately when HOME has not been set, I didn’t notice cproto looking for HOME though. The memory system is different on RISC OS 4 so I can’t really compare it exactly to my Iyonix 5.20. Maybe someone else with RISC OS 4 out there could try the program?
I have used gcc 4.1.2 on a RiscPc RISC OS 3.7 but I always had at least 32MB of
memory on those machines.

Edit: I have now compiled cproto with gcc3.4 for use with gcc3.4 binaries.
Check the Boot file for the include directories setting
The result is much better than I got with 4.1.2 for some reason.

hi
the last version you have compiled is working without any crash. Unfortunately it has been configured to remove the C prototypes in the proper template

i mean, if you have a source.c with

int foo(int a)
{
}

cproto source.c should print

int foo(int a);

actually it prints

foo

but, hey ? got progress =D

cproto source.c should print int foo(int a);
actually it prints foo

Yes, that was the output from using cc1 directly, I have changed it to use the cpp within !gcc 3.4 now.
If the cpp is not turned off it outputs lots. So to output just the bare functions(prototype?) you need

cproto -qE 0 acorn.c
/* acorn.c */
void detect_adfs(SECTION *section, int level);

here you can see a solution for VIM; it runs as a GUI task, as an ANSI task with Nettle which supports most common terminal types, including vt100, vt102, vt220 and xterm terminal emulation in full color.

might the trick also be fine for text-editor/nano ?