There is no A_UNDERLINE effect in tmux on Mac. This problem reproduces below.
https://gist.github.com/ta1kt0me/d890ef0b25513be40e2e1e3b60d0d51a

• OS: Mac OS Sierra
• Ruby: 2.4.0p0
• tmux: 2.3
• curses: 1.2.0

note

If don't use tmux, enable A_UNDERLINE effect.

• not use tmux

• use tmux

When running gem install curses on Ubuntu 19.04 building the native extension for curses v1.3.1 fails for Ruby versions <=2.4.4. libncursesw5-dev and libncurses5-dev are installed. I tried with C-Ruby versions installed via RVM.

Because the failure logs are about 4900 lines long, I’ll attach them as a file:
curses-1.3.1.zip

Installing the gem works for me for Ruby >=2.4.5.

Hi!

Please try the following program:

#!/usr/bin/ruby

require 'curses'
include Curses

init_screen
crmode

setpos((lines - 5) / 2, (cols - 10) / 2)
addstr('Hit any key')
refresh

getch

# Don't close_screen(); we want the screen contents to remain after
# exiting

Expected result:

• After you hit some key, the program should quit and the "Hit any key" message should still be visible on screen

Current result:

• When the program exits, the screen contents is restored to what it was before the program was started

Note:

• If you kill -9 the program, it exits without restoring the screen contents just like I want, but then it exits with a non-zero return code.

The use case is that I want my pager implementation to be able to exit without restoring screen contents when pressing ^C:
https://github.com/walles/moar

I currently use the kill -9 workaround, but would like to be able to do without it.

Regards /Johan

Note that this issue is cloned from https://bugs.ruby-lang.org/issues/9177 upon request from Hiroshi SHIBATA.

I noticed that on Windows the curses gem can be currently installed with just gem install curses.
Why is this not clearly written in the readme?

I was under the impression that a binary was not available for windows so I lost a lot of time looking for other ways to install it!

It should be more clear that on Windows all you have to do is "gem install curses".

I'm using OpenSUSE 15.2 OS. And I'm running ruby 3.0,3 installed over rbenv and ruby 2.5.0 on real machine.
I had run all samples on my laptop without problems but when I run sample/colors.rb I get this error:

ruby colors.rb
colors.rb:14:in colors': colors() function is unimplemented on this machine (NotImplementedError) from colors.rb:14:in

'

Could you give me any suggestion to fix it? Thanks!

Is there a way to output text to a curses window without curses messing with the ansi color codes - I would like the color codes to simply pass through and let the text be colored in the window. I am perfectly aware of how to color text in curses, but I would rather not go through that roundabout when outputting syntax highlighted text to a curses window (like with the command

curses_window << `batcat --color=always #{file}`

... where batcat is the syntax highlighting cousin to cat (also known as just "bat")

Please add one or two sentences more about why it is not longer part, since when and who and where that has been decided. Many thanks.

Trying to install the gem as per documentation. Tried Rubies 3.3, 3.2, 3.1.

$ gem install curses -- --use-system-libraries --with-ncurses-dir=/opt/homebrew/opt/ncurses
Fetching curses-1.4.4.gem
Building native extensions with: '--use-system-libraries --with-ncurses-dir=/opt/homebrew/opt/ncurses'
This could take a while...
ERROR:  Error installing curses:
	ERROR: Failed to build gem native extension.

    current directory: /Users/piotr/.gem/ruby/3.1.3/gems/curses-1.4.4/ext/curses
/Users/piotr/.rubies/ruby-3.1.3/bin/ruby -I /Users/piotr/.rubies/ruby-3.1.3/lib/ruby/3.1.0 extconf.rb --use-system-libraries --with-ncurses-dir\=/opt/homebrew/opt/ncurses
checking for tgetent() in -ltinfow... no
checking for tgetent() in -ltinfo... no
checking for tgetent() in -ltermcap... yes
checking for ncursesw/curses.h... yes
checking for initscr() in -lncursesw... yes
header: ncursesw/curses.h
library: ncursesw
checking for beep()... yes
checking for bkgd()... yes
checking for bkgdset()... yes
checking for curs_set()... yes
checking for deleteln()... yes
checking for doupdate()... yes
checking for flash()... yes
checking for getbkgd()... yes
checking for getnstr()... yes
checking for init()... no
checking for init in ncursesw/curses.h... no
checking for isendwin()... yes
checking for keyname()... yes
checking for keypad()... yes
checking for resizeterm()... yes
checking for scrl()... yes
checking for set()... no
checking for set in ncursesw/curses.h... no
checking for setscrreg()... yes
checking for ungetch()... yes
checking for addnwstr()... yes
checking for wattroff()... yes
checking for wattron()... yes
checking for wattrset()... yes
checking for wbkgd()... yes
checking for wbkgdset()... yes
checking for wdeleteln()... yes
checking for wgetnstr()... yes
checking for wresize()... yes
checking for wscrl()... yes
checking for wsetscrreg()... yes
checking for werase()... yes
checking for redrawwin()... yes
checking for waddnwstr()... yes
checking for mvderwin()... yes
checking for derwin()... yes
checking for touchwin()... yes
checking for untouchwin()... yes
checking for wtouchln()... yes
checking for is_linetouched()... yes
checking for is_wintouched()... yes
checking for def_prog_mode()... yes
checking for reset_prog_mode()... yes
checking for timeout()... yes
checking for wtimeout()... yes
checking for nodelay()... yes
checking for init_color()... yes
checking for wcolor_set()... yes
checking for use_default_colors()... yes
checking for assume_default_colors()... yes
checking for newpad()... yes
checking for unget_wch()... yes
checking for get_wch()... yes
checking for wget_wch()... yes
checking for PDC_get_key_modifiers()... no
checking for PDC_get_key_modifiers in ncursesw/curses.h... no
checking for convertible type of chtype... unsigned int
checking for _XOPEN_SOURCE_EXTENDED... yes
checking for ESCDELAY in ncursesw/curses.h... yes
checking for TABSIZE in ncursesw/curses.h... yes
checking for COLORS in ncursesw/curses.h... yes
checking for COLOR_PAIRS in ncursesw/curses.h... yes
checking for function curses_version in ncursesw/curses.h... yes
checking for variable curses_version in ncursesw/curses.h... no
checking for ncursesw/menu.h... yes
checking for new_menu() in -lmenuw... yes
checking for ncursesw/form.h... yes
checking for new_form() in -lformw... yes
checking for form_driver_w()... yes
checking for sizeof(WINDOW) is available... yes
checking for sizeof(MEVENT) is available... yes
checking for sizeof(ITEM) is available... yes
checking for sizeof(MENU) is available... yes
checking for sizeof(FIELD) is available... yes
checking for sizeof(FORM) is available... yes
creating extconf.h
creating Makefile

current directory: /Users/piotr/.gem/ruby/3.1.3/gems/curses-1.4.4/ext/curses
make DESTDIR\= sitearchdir\=./.gem.20240313-99458-tl51wo sitelibdir\=./.gem.20240313-99458-tl51wo clean

current directory: /Users/piotr/.gem/ruby/3.1.3/gems/curses-1.4.4/ext/curses
make DESTDIR\= sitearchdir\=./.gem.20240313-99458-tl51wo sitelibdir\=./.gem.20240313-99458-tl51wo
compiling curses.c
curses.c:3072:1: warning: function 'no_item' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
{
^
curses.c:3270:1: warning: function 'no_menu' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
{
^
curses.c:5103:5: error: incompatible function pointer types passing 'VALUE (VALUE, VALUE)' (aka 'unsigned long (unsigned long, unsigned long)') to parameter of type 'VALUE (*)(VALUE)' (aka 'unsigned long (*)(unsigned long)') [-Wincompatible-function-pointer-types]
    rb_define_method(cItem, "opts", item_opts_m, 0);
    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:287:135: note: expanded from macro 'rb_define_method'
#define rb_define_method(klass, mid, func, arity)           RBIMPL_ANYARGS_DISPATCH_rb_define_method((arity), (func))((klass), (mid), (func), (arity))
                                                                                                                                      ^~~~~~
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:276:1: note: passing argument to parameter here
RBIMPL_ANYARGS_DECL(rb_define_method, VALUE, const char *)
^
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:254:72: note: expanded from macro 'RBIMPL_ANYARGS_DECL'
RBIMPL_ANYARGS_ATTRSET(sym) static void sym ## _00(__VA_ARGS__, VALUE(*)(VALUE), int); \
                                                                       ^
curses.c:5124:5: error: incompatible function pointer types passing 'VALUE (VALUE, VALUE)' (aka 'unsigned long (unsigned long, unsigned long)') to parameter of type 'VALUE (*)(VALUE)' (aka 'unsigned long (*)(unsigned long)') [-Wincompatible-function-pointer-types]
    rb_define_method(cMenu, "fore", menu_get_fore, 0);
    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:287:135: note: expanded from macro 'rb_define_method'
#define rb_define_method(klass, mid, func, arity)           RBIMPL_ANYARGS_DISPATCH_rb_define_method((arity), (func))((klass), (mid), (func), (arity))
                                                                                                                                      ^~~~~~
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:276:1: note: passing argument to parameter here
RBIMPL_ANYARGS_DECL(rb_define_method, VALUE, const char *)
^
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:254:72: note: expanded from macro 'RBIMPL_ANYARGS_DECL'
RBIMPL_ANYARGS_ATTRSET(sym) static void sym ## _00(__VA_ARGS__, VALUE(*)(VALUE), int); \
                                                                       ^
curses.c:5126:5: error: incompatible function pointer types passing 'VALUE (VALUE, VALUE)' (aka 'unsigned long (unsigned long, unsigned long)') to parameter of type 'VALUE (*)(VALUE)' (aka 'unsigned long (*)(unsigned long)') [-Wincompatible-function-pointer-types]
    rb_define_method(cMenu, "back", menu_get_back, 0);
    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:287:135: note: expanded from macro 'rb_define_method'
#define rb_define_method(klass, mid, func, arity)           RBIMPL_ANYARGS_DISPATCH_rb_define_method((arity), (func))((klass), (mid), (func), (arity))
                                                                                                                                      ^~~~~~
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:276:1: note: passing argument to parameter here
RBIMPL_ANYARGS_DECL(rb_define_method, VALUE, const char *)
^
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:254:72: note: expanded from macro 'RBIMPL_ANYARGS_DECL'
RBIMPL_ANYARGS_ATTRSET(sym) static void sym ## _00(__VA_ARGS__, VALUE(*)(VALUE), int); \
                                                                       ^
curses.c:5130:5: error: incompatible function pointer types passing 'VALUE (VALUE, VALUE)' (aka 'unsigned long (unsigned long, unsigned long)') to parameter of type 'VALUE (*)(VALUE)' (aka 'unsigned long (*)(unsigned long)') [-Wincompatible-function-pointer-types]
    rb_define_method(cMenu, "opts", menu_opts_m, 0);
    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:287:135: note: expanded from macro 'rb_define_method'
#define rb_define_method(klass, mid, func, arity)           RBIMPL_ANYARGS_DISPATCH_rb_define_method((arity), (func))((klass), (mid), (func), (arity))
                                                                                                                                      ^~~~~~
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:276:1: note: passing argument to parameter here
RBIMPL_ANYARGS_DECL(rb_define_method, VALUE, const char *)
^
/Users/piotr/.rubies/ruby-3.1.3/include/ruby-3.1.0/ruby/internal/anyargs.h:254:72: note: expanded from macro 'RBIMPL_ANYARGS_DECL'
RBIMPL_ANYARGS_ATTRSET(sym) static void sym ## _00(__VA_ARGS__, VALUE(*)(VALUE), int); \
                                                                       ^
2 warnings and 4 errors generated.
make: *** [curses.o] Error 1

make failed, exit code 2

Gem files will remain installed in /Users/piotr/.gem/ruby/3.1.3/gems/curses-1.4.4 for inspection.
Results logged to /Users/piotr/.gem/ruby/3.1.3/extensions/arm64-darwin-22/3.1.0-static/curses-1.4.4/gem_make.out

I tried this:

gem install ncurses

The output was that:

"Fetching ncurses-0.9.1.gem
ERROR: Error installing ncurses:
#<Gem::Specification name=ncurses version=0.9.1> has an invalid specification_version
"

So first I thought this was the curses gem here, but I believe this is another gem.

Perhaps the main README could mention this? I was about to report the above but then
I realized that the name is "curses" rather than "ncurses". A bit confusing to me; I made
the incorrect assumption that ncurses would be curses and vice versa.

Despite:

$ sudo apt install libncursesw5 libncursesw5-dev

window.box(' ', "\u2500")

outputs a horizontal block of gray text instead of a nice horizontal box line.

What steps are needed to display wide unicode characters using the curses gem?

I can confirm the version of ruby does properly output the unicode I expect:

 puts "\u2500"*9
─────────

https://bugs.python.org/issue36982

"A new function called curses.has_extended_color_support() will indicate whether the linked ncurses library provides extended color support. It returns true if curses.h defines NCURSES_EXT_COLORS and NCURSES_EXT_FUNCS, indicating that the extended color functions are available. This seems more useful to developers than using an indirect method like trying to set a color-pair greater than 0x7fff and checking for an exception to indicate lack of support."

Would this make sense to add for ruby too? E. g. a method such as:

Curses.has_extended_color_support?

I'd love to be able to use more than 8 colours via ncurses. In particular the html-colours:

https://htmlcolorcodes.com/colors/

I could then add this to my gem e. g.

CursesParadise.steelblue("Hello world!\n")

As easier alternative.

After a gem update, I'm suddenly seeing this build failure:

current directory: /opt/slangs/rb/mygems/gems/curses-1.3.0/ext/curses
make "DESTDIR="
compiling curses.c
curses.c:218:1: warning: function 'no_window' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
{
^
curses.c:244:63: error: invalid application of 'sizeof' to an incomplete type 'WINDOW' (aka 'struct _win_st')
    if (winp->window && winp->window != stdscr) size += sizeof(*winp->window);
                                                              ^~~~~~~~~~~~~~~
/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.14.sdk/usr/include/curses.h:322:16: note: forward declaration of 'struct _win_st'
typedef struct _win_st WINDOW;
               ^
1 warning and 1 error generated.
make: *** [curses.o] Error 1

make failed, exit code 2

This is on macOS 10.14.4. with Xcode 10.2, and the operative ruby is version 2.6.2 that I built and installed myself. The previous versions of the curses gem built just fine.

I just installed the gem.

Response was:

Fetching curses-1.4.4.gem
Building native extensions. This could take a while...
Successfully installed curses-1.4.4
Parsing documentation for curses-1.4.4
Couldn't find file to include 'sample/hello.rb' from ext/curses/curses.c
Couldn't find file to include 'sample/rain.rb' from ext/curses/curses.c
Couldn't find file to include 'sample/mouse.rb' from ext/curses/curses.c
Installing ri documentation for curses-1.4.4
Done installing documentation for curses after 0 seconds
1 gem installed

The three warnings relate to some .rb file in sample - see lines starting
from 5-7.

Could these three .rb files be included? hello.rb rain.rb and mouse.rb. Not
sure why they could not be found.

So I just installed the curses-1.4.1 gem.

However had, in the sample/ directory I see the following 7 .rb files:

form.rb
hello.rb
menu.rb
mouse.rb
rain.rb
view2.rb
view.rb

On github, though, if you go here:

https://github.com/ruby/curses/tree/master/sample

You can see 11 files:

addch.rb
attr_demo.rb
colors.rb
form.rb
hello.rb
menu.rb
mouse.rb
mouse_move.rb
rain.rb
view.rb
view2.rb 

I am quite confused. Is the gem built from some other source target that explains why 4 files seem to be missing? Or is this some user error?

I am relearning curses (have had old code with Ncurses as toplevel module name) so I randomly looked at attr_demo.rb
first via the browser. Then I wanted a local copy to experiment, and just downloaded the gem, then noticed that some files are missing. Very strange..

The documentation says that addch would support outputting characters with attributes, but in practice it doesn't work.

Ways to reproduce:

require "curses"
include Curses

init_screen
begin
  addstr("The following letter A should be BOLD and UNDERLINED by using addch:\n")
  addch('A'.ord | A_BOLD | A_UNDERLINE)

  addstr("\nIt should look the same as when using attron and addstr:\n")
  attron(A_BOLD | A_UNDERLINE)
  addstr("A")
  getch
ensure
  close_screen
end

What I am seeing:

Is just the documentation inaccurate or does the conversion from NUM2CH remove the attributes again?

Hi,

I've been attempting to get curses working for ruby on my machine, but the first init_screen crashes everything.

stack trace:
log.txt

I've narrowed it down to libtinfo, and bringing in libtinfow.so fixes it.

Not working version:

$ ldd ~/.gem/ruby/2.4.0/gems/curses-1.3.1/lib/curse
	linux-vdso.so.1 (0x00007ffd47b59000)
	libruby24.so.2.4 => /usr/lib64/libruby24.so.2.4 (0x00007f736b160000)
	libformw.so.6 => /usr/lib64/libformw.so.6 (0x00007f736b14c000)
	libmenuw.so.6 => /usr/lib64/libmenuw.so.6 (0x00007f736b142000)
	libncursesw.so.6 => /lib64/libncursesw.so.6 (0x00007f736b108000)
	libtinfo.so.6 => /lib64/libtinfo.so.6 (0x00007f736b0cc000)
	libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f736b0aa000)
	libc.so.6 => /lib64/libc.so.6 (0x00007f736aed9000)
	libgmp.so.10 => /usr/lib64/libgmp.so.10 (0x00007f736ae60000)
	libdl.so.2 => /lib64/libdl.so.2 (0x00007f736ae5a000)
	libcrypt.so.1 => /lib64/libcrypt.so.1 (0x00007f736ae21000)
	libm.so.6 => /lib64/libm.so.6 (0x00007f736ace3000)
	/lib64/ld-linux-x86-64.so.2 (0x00007f736b453000)
	libtinfow.so.6 => /lib64/libtinfow.so.6 (0x00007f736aca7000)

Notice how libtinfow is being pulled in, just after libtinfo

Here's the quick patch I made to get everything up and running:
tinfo.patch.gz

And the linked libraries after, works:

$ ldd ~/.gem/ruby/2.4.0/gems/curses-1.3.1/lib/curse
	linux-vdso.so.1 (0x00007ffe2554b000)
	libruby24.so.2.4 => /usr/lib64/libruby24.so.2.4 (0x00007f2ffaad9000)
	libformw.so.6 => /usr/lib64/libformw.so.6 (0x00007f2ffaac5000)
	libmenuw.so.6 => /usr/lib64/libmenuw.so.6 (0x00007f2ffaabb000)
	libncursesw.so.6 => /lib64/libncursesw.so.6 (0x00007f2ffaa81000)
	libtinfow.so.6 => /lib64/libtinfow.so.6 (0x00007f2ffaa45000)
	libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f2ffaa23000)
	libc.so.6 => /lib64/libc.so.6 (0x00007f2ffa852000)
	libgmp.so.10 => /usr/lib64/libgmp.so.10 (0x00007f2ffa7d9000)
	libdl.so.2 => /lib64/libdl.so.2 (0x00007f2ffa7d3000)
	libcrypt.so.1 => /lib64/libcrypt.so.1 (0x00007f2ffa79a000)
	libm.so.6 => /lib64/libm.so.6 (0x00007f2ffa65c000)

I suspect the cause of this is in part because of the distro I'm running, gentoo, which allows me to compile libraries in all sorts of wacky ways.

To aid with debugging, here's my install log for ncurses:
ncurses_install.log.gz

Is there any support for modifying attributes for characters already on the screen in Ruby curses? It seems that the chgat family of functions aren't supported. Since they are particularly helpful for rendering virtual cursors, I was wondering how this is solved in Ruby curses.

I'm trying to run the example at https://github.com/ruby/curses/blob/master/sample/mouse.rb and I'm getting this error:

test-curses.rb:29:in <main>: uninitialized constant BUTTON1_CLICKED (NameError)

I'm on Ruby 2.6.5p114 (2019-10-01 revision 67812) [x64-mingw32] running on Windows 10.

I tried to install Curses while using Ruby 2.7 on Windows, but it seems 2.7 is not allowed yet.

      curses (~> 1.3.2) x64-mingw32 was resolved to 1.3.2, which depends on
        Ruby (< 2.7.dev, >= 2.5) x64-mingw32

Could you please take a look at it if it's possible to change the requirement?

Hi.

The .box() method does not work with UTF-8 characters.

window.box('│', '─') leads to displaying â character (it is grey rectangle in my terminal).

Where these dashes are box-drawing characters copied directly into the code (with unicode transcription it doesn't work either).

What works however is UTF-8, but with .addstr() method.

Curses 1.2.4
Ruby 2.4
Fedora 25

I have been stuck using a very old library and unable to switch to this Curses library because of the absence of newterm. I do curses over telnet with multiple screens simultaneously so this method is critical for me.

Any chance you can add this, I'm at the point where I'm about to fork this project as the old library is crashing on me.

I used Ruby 1.9 x86 with Windows. Now I installed 2.1 x64. I discovered that curses is not bundle anymore with Ruby and this gem is supposed to be the replacement. A quick gem install curses failed, because some native libraries needed to be built. I installed the dev-kit. It does not seem to include the required headers. I might be able to get the headers from cygwin by installing ``libncurses-devel`, but that is already quite a lot of effort already after installing the devkit.

Is there some way to have the gem include a prebuilt dll for Windows? For Python, it seems to be possible to offer packages with binaries for platforms to not require compiling the native libraries. As it currently stands, it is quite difficult to just get up and running if the devkit is not sufficient for compiling the gem.

It used to be really easy for a Windows user to use ncurses with Ruby since it was bundled with the Ruby install. The gem is not a sufficient replacement on Windows as a platform, if there are not easy instructions and an easy way to build to gem. On Linux it would be quite simple to obtain the missing headers and build the gem, but this is not the case for Windows.

I downloaded curses using "gem install curses", while using ruby 2.3.0 whenever i attempt to start a script in which i used curses i get "

C:/RailsInstaller/Ruby2.3.0/lib/ruby/2.3.0/rubygems/core_ext/kernel_require.rb:5
5:in require': cannot load such file -- curses.so (LoadError) from C:/RailsInstaller/Ruby2.3.0/lib/ruby/2.3.0/rubygems/core_ext/kernel _require.rb:55:in require'
from C:/RailsInstaller/Ruby2.3.0/lib/ruby/gems/2.3.0/gems/curses-1.2.3-x
86-mingw32/lib/curses.rb:17:in rescue in <top (required)>' from C:/RailsInstaller/Ruby2.3.0/lib/ruby/gems/2.3.0/gems/curses-1.2.3-x 86-mingw32/lib/curses.rb:13:in <top (required)>'
from C:/RailsInstaller/Ruby2.3.0/lib/ruby/2.3.0/rubygems/core_ext/kernel
_require.rb:127:in require' from C:/RailsInstaller/Ruby2.3.0/lib/ruby/2.3.0/rubygems/core_ext/kernel _require.rb:127:in rescue in require'
from C:/RailsInstaller/Ruby2.3.0/lib/ruby/2.3.0/rubygems/core_ext/kernel
_require.rb:40:in require' from test.rb:1:in

rake newb fails because I don't have hoe installed on my machine, but it's required in the Rakefile. This should probably be added there in the README.

Hi,

I've been trying the different sample code for curses. All of them work except for the sample/menu.rb
Running ruby sample/menu.rb gives me this error.

sample/menu.rb:13:in `<main>': uninitialized constant Curses::Menu (NameError)

menu = Curses::Menu.new([

I'm on mac OS ventura 13.4.1 and homebrew seems to have ncurses 6.4 already installed.

I can see that there are some issues getting menu from ncurses from MacOS

• https://www.linuxquestions.org/questions/programming-9/on-macos-10-15-configure-script-can%27t-find-ncurses-menu-library-4175695096/
• https://stackoverflow.com/questions/56622042/clang-on-macos-fails-linking-lmenu-from-ncurses

Is this why there is no documentation on the README for MacOS?

The menu sample works fine when inside the official docker image ruby:3.0

What would it take to get ncurses menu to work?

Thanks for your help

This is basically what's been reported here https://bugs.ruby-lang.org/issues/9094

In my case I'm getting 195 and 169 for é, but really it's exactly the same issue like Johan described in the link above.

Cheers,

J

I can't capture any of the multibyte characters (ie. function keys, arrow keys, and ESC) because they are all evaluating to the same thing: "\e".

Description of Curses.get_char: "Read and returns a character or function key from the window." (emphasis mine)

Single byte characters work fine, and escape code characters (\t for tab) work.

Using version 1.4.3
(I'd like to navigate through text buffers with directional keys)

Thank you for any insight or feedback in advance.

Hi,

I'm getting this error when doing gem install curses:

error: unknown warning option '-Wno-cast-function-type'; did you mean '-Wno-bad-function-cast'? [-Werror,-Wunknown-warning-option]

Apple clang version 13.0.0 (clang-1300.0.27.3)
Target: arm64-apple-darwin21.3.0
Thread model: posix
InstalledDir: /Library/Developer/CommandLineTools/usr/bin

Attached is a simple curses program that displays some text and calls getch. It also sets up a trap on SIGWINCH so that it can detect window size changes.

require 'curses'
include Curses

@listeners = []

def size_changed
  # TODO -- notify all size changed listeners
  @listeners.each { |listener|
    #listener.screen_size_changed
 }
end


begin
  Signal.trap('SIGWINCH', size_changed )

  init_screen
  Curses.raw
  clear
  curs_set(0)
  noecho
  cbreak
  start_color

  Curses.stdscr.setpos(0, 0)
  Curses.stdscr.addstr("Press any key ")
  Curses.stdscr.getch

rescue SystemExit, Interrupt
  # Empty Catch block so ruby doesn't puke out
  # a stack trace when CTRL-C is used
ensure
  Curses.close_screen
end

The issue is that the call to an array method causes a NoMethodError when the size_changed method gets called.

To reproduce:

1. Run the sample program
2. Resize the terminal window

I also found this post:
http://masanjin.net/blog/ruby-ncurses-and-thread-blocking

That talks about some implementations of curses blocking threads on getch.

I took a look at the ncurses-ruby gem and found some interesting code in C, entitled rbncurshelper_nonblocking_wgetch

https://github.com/eclubb/ncurses-ruby/blob/master/ext/ncurses/ncurses_wrap.c#L792

Basically, I'm writing a GUI library on top of curses and would like to handle window resizes by notifying all objects to redraw themselves. Was wondering.

1. Are there limitations to what you can do when handling a signal?
2. Is getch blocking?
3. Is there a better way to handle this then catching NoMethodError's like below

 def get_ch_handle_signals
      got_input = false
      while !got_input
        begin
          c = getch
          got_input = true
        rescue NoMethodError
          # TODO -- SIGWINCH could have occurred -- reposition..
          c = ''
        end
      end

      c
    end

It seems various constants are missing from the documentation at

https://www.rubydoc.info/gems/curses

For instance A_TOP (top highlight) is not listed (and can't be found using the search).

I'm, currently experiencing an issue with the "inch" method in Ruby curses, this is reproducible in both curses 1.4.2 and 1.4.1. However the issue is not present in curses 1.4.0

Using win.inch.chr returns the character at the cursor position as intended, but in curses 1.4.1/2 this results in an error saying char is out of range. Manually setting the encoding to UTF_8 or UTF_16 removes the error message but does not return the character as intended

ie
Curses 1.4.0 win.inch.chr expected behaviour
Curses 1.4.1+ win.inch.chr error, char out of range
Curses 1.4.1+ win.inch.chr(Encoding::UTF_8) returns ? block
Curses 1.4.1+ win.inch.chr(Encoding::UTF_16) interferes with formatting

I was wondering, is it intended that this gem work on Ruby 1.9.3-p484 ? I'm aware that it's an extraction of curses from the Ruby standard library (https://bugs.ruby-lang.org/issues/8584), being removed from Ruby 2.1.0 onwards, and that Ruby 1.9.3 already has curses included. On Ruby 2.0.0-p353 it installs fine for me, however, despite curses also being included in that standard library. The problem is with trying to use curses within a separate gem whilst supporting Ruby 1.9.3, 2.0.0, 2.1.0 for that gem. With the Ruby 1.9.3 gem install failure, I can't specify curses as a formal dependency within a gemspec—yet it's needed for the Ruby 2.1.0 support. I've got around it, but I'm interested to learn whether support for Ruby 1.9.3 is intended, and whether perhaps I've missed something. :)

I'm attempting to install on OS X 10.9.1 (13B42) with Xcode 5.0.2 (5A3005), RVM 1.25.14 (master), Gem 2.2.0, Bundler 1.5.1. On Ruby 2.1.0-p0 all seems well with curses gem installed, on Ruby 2.0.0-p353 gem installs without error, on Ruby 1.9.3-p484 I get the following:

$ gem install curses -V
HEAD https://rubygems.org/latest_specs.4.8.gz
302 Moved Temporarily
HEAD https://s3.amazonaws.com/production.s3.rubygems.org/latest_specs.4.8.gz
200 OK
GET https://rubygems.org/latest_specs.4.8.gz
302 Moved Temporarily
GET https://s3.amazonaws.com/production.s3.rubygems.org/latest_specs.4.8.gz
200 OK
HEAD https://rubygems.org/specs.4.8.gz
302 Moved Temporarily
HEAD https://s3.amazonaws.com/production.s3.rubygems.org/specs.4.8.gz
200 OK
GET https://rubygems.org/specs.4.8.gz
302 Moved Temporarily
GET https://s3.amazonaws.com/production.s3.rubygems.org/specs.4.8.gz
200 OK
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/BSDL
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/COPYING
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/History.md
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/Manifest.txt
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/README.md
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/Rakefile
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/curses.gemspec
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/ext/curses/curses.c
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/ext/curses/depend
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/ext/curses/extconf.rb
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/lib/curses.rb
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/sample/hello.rb
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/sample/mouse.rb
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/sample/rain.rb
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/sample/view.rb
/Users/mlnw/.rvm/gems/ruby-1.9.3-p484/gems/curses-1.0.0/sample/view2.rb
Building native extensions. This could take a while...
/Users/mlnw/.rvm/rubies/ruby-1.9.3-p484/bin/ruby extconf.rb
checking for tgetent() in -ltinfo... no
checking for tgetent() in -ltermcap... yes
checking for ncurses.h... yes
checking for initscr() in -lncursesw... no
checking for initscr() in -lncurses... yes
header: ncurses.h
library: ncurses
checking for beep()... yes
checking for bkgd()... yes
checking for bkgdset()... yes
checking for curs_set()... yes
checking for deleteln()... yes
checking for doupdate()... yes
checking for flash()... yes
checking for getbkgd()... yes
checking for getnstr()... yes
checking for init()... no
checking for init in ncurses.h... no
checking for isendwin()... yes
checking for keyname()... yes
checking for keypad()... yes
checking for resizeterm()... yes
checking for scrl()... yes
checking for set()... no
checking for set in ncurses.h... no
checking for setscrreg()... yes
checking for ungetch()... yes
checking for wattroff()... yes
checking for wattron()... yes
checking for wattrset()... yes
checking for wbkgd()... yes
checking for wbkgdset()... yes
checking for wdeleteln()... yes
checking for wgetnstr()... yes
checking for wresize()... yes
checking for wscrl()... yes
checking for wsetscrreg()... yes
checking for def_prog_mode()... yes
checking for reset_prog_mode()... yes
checking for timeout()... yes
checking for wtimeout()... yes
checking for nodelay()... yes
checking for init_color()... yes
checking for wcolor_set()... yes
checking for use_default_colors()... yes
checking for newpad()... yes
checking for ESCDELAY in ncurses.h... yes
checking for TABSIZE in ncurses.h... yes
checking for COLORS in ncurses.h... yes
checking for COLOR_PAIRS in ncurses.h... yes
checking for function curses_version in ncurses.h... yes
checking for variable curses_version in ncurses.h... no
creating Makefile
make
make: *** No rule to make target /Users/mlnw/.rvm/rubies/ruby-1.9.3-p484/include/ruby-1.9.1/ruby/thread.h', needed bycurses.o'. Stop.
ERROR: Error installing curses:
ERROR: Failed to build gem native extension.

Building has failed. See above output for more information on the failure.

Peace,
tiredpixel

If the code displays non-US-ASCII characters they do not appear as expected.

Add some accents to the sample/hello.rb example, like

  show_message("Hello, World!: éàù")

The message is displayed as:

                        ┌----------------------┐
                        |                      |
                        |  Hello, World!: M-CM-)
                        M-CM- M-CM-9           |
                        └----------------------┘

Hi, on mingw Ruby26-x64, on Windows 7, using curses-1.4.4 and the statically-linked default PDCurses, attron(color_pair(n)) doesn't have any effect, and I only get the default gray-on-black, from this example:

require 'curses'
include Curses
init_screen
start_color
init_pair(6, COLOR_BLUE, COLOR_RED)
attron(color_pair(6)) do
  setpos(5,5)
  addstr('this should be in blue on red')
end
getch
close_screen

It looks like PDCurses is compiled in 64-bit chtype mode by default, and the color bits are shifting off the end of a regular 32-bit long and getting truncated in window_attron(), window_attroff(), window_attrset(), and curses_color_pair() by the INT2FIX()/NUM2INT() macros.

Replacing these macros with NUM2CHTYPE() and CHTYPE2NUM() in the color/attr functions seems to resolve the problem and the example above works as expected.

Hi,

This has been bugging me for a while, and I do not know if I am doing something wrong, or if it is a bug in either curses, ncurses or in my terminal.

I have created a fairly elaborate script that first loads in a colorscheme, I have stolen the material colors scheme from Android to work with (please see the material-colors.json file here: https://gist.github.com/kaspergrubbe/f6e64178e3962a3609e93569dc7722ce). It then reads all the hex values, and converts it to ncurses (between 0 and 1000) like this:

@red = 'ff'
@green = 'ff'
@blue = 'ff'
[@red, @green, @blue].map{|h| h.hex / 0.255}.map(&:to_i)
#=> [1000, 1000, 1000]

I then read the materials-colors.json file, it is grouped by colorname and a weight like so:

{
  "red": {
    "50": "#ffebee",
    "100": "#ffcdd2",
    "300": "#e57373",
    "500": "#f44336",
    "700": "#d32f2f",
    "800": "#c62828",
    "900": "#b71c1c"
  },
  "pink": {
    "50": "#fce4ec",
    "100": "#f8bbd0",
    "300": "#f06292",
    "500": "#e91e63",
snipped

I then load in all the individual colours, calculate all possible combination of colour pairs where each color can be both front and back, and then I initialise curses and get it to setup the colours.

I then print lol 5 times:

  fg = @colour_map['grey']['50']
  bg = @colour_map['lightblue']['900']
  pair = @colour_map.pair(fg, bg)

  paint_effect(win, Curses.color_pair(pair.number) | Curses::A_UNDERLINE) do
    win.setpos(1,1)
    win.addstr('lol')
  end

  paint_effect(win, Curses.color_pair(1337) | Curses::A_UNDERLINE) do
    win.setpos(2,2)
    win.addstr('lol')
  end

  paint_effect(win, Curses.color_pair(10337) | Curses::A_UNDERLINE) do
    win.setpos(3,3)
    win.addstr('lol')
  end

  paint_effect(win, Curses.color_pair(0)) do
    win.setpos(4,4)
    win.addstr('lol')
  end

  paint_effect(win, Curses.color_pair(@colour_map.pair(@colour_map['grey']['50'], @colour_map['blue']['900']).number)) do
    win.setpos(5,5)
    win.addstr('lol')
  end

But the results are very strange. This is how iTerm renders it:

If you hover over it, you can see that it writes some text:

And the OSX Terminal.app renders it very differently:

However, while Terminal.app manages to reset the colors after the program exits, the same cannot be said about iTerm:

Before:

After:

I have made a whole repo here: https://github.com/kaspergrubbe/curses_colortest the only thing to test is just cloning the repo and running ruby colortest.rb.

Can you see something I can't?

Maybe by design, anyway, someone may want to have a look at it:

gem install curses -- --with-ncurses-dir=$(brew --prefix ncurses)

But 1.2.7 can install directly.

When installing curses the extconf.rb configuration crashes out on checking for convertible type of chtype:

checking for tgetent() in -ltinfo... yes
checking for ncursesw/curses.h... yes
checking for initscr() in -lncursesw... yes
header: ncursesw/curses.h
library: ncursesw
checking for beep()... yes
checking for bkgd()... yes
checking for bkgdset()... yes
checking for curs_set()... yes
checking for deleteln()... yes
checking for doupdate()... yes
checking for flash()... yes
checking for getbkgd()... yes
checking for getnstr()... yes
checking for init()... no
checking for init in ncursesw/curses.h... no
checking for isendwin()... yes
checking for keyname()... yes
checking for keypad()... yes
checking for resizeterm()... yes
checking for scrl()... yes
checking for set()... no
checking for set in ncursesw/curses.h... no
checking for setscrreg()... yes
checking for ungetch()... yes
checking for addnwstr()... yes
checking for wattroff()... yes
checking for wattron()... yes
checking for wattrset()... yes
checking for wbkgd()... yes
checking for wbkgdset()... yes
checking for wdeleteln()... yes
checking for wgetnstr()... yes
checking for wresize()... yes
checking for wscrl()... yes
checking for wsetscrreg()... yes
checking for werase()... yes
checking for redrawwin()... yes
checking for waddnwstr()... yes
checking for touchwin()... yes
checking for untouchwin()... yes
checking for wtouchln()... yes
checking for is_linetouched()... yes
checking for is_wintouched()... yes
checking for def_prog_mode()... yes
checking for reset_prog_mode()... yes
checking for timeout()... yes
checking for wtimeout()... yes
checking for nodelay()... yes
checking for init_color()... yes
checking for wcolor_set()... yes
checking for use_default_colors()... yes
checking for assume_default_colors()... yes
checking for newpad()... yes
checking for unget_wch()... yes
checking for get_wch()... yes
checking for wget_wch()... yes
checking for PDC_get_key_modifiers()... no
checking for PDC_get_key_modifiers in ncursesw/curses.h... no
checking for convertible type of chtype... failed
*** ext/curses/extconf.rb failed ***
Could not create Makefile due to some reason, probably lack of necessary
libraries and/or headers.  Check the mkmf.log file for more details.  You may
need configuration options.

This is with RVM-built Ruby 2.4.1 on both Fedora 28 (ncurses 6.1) and Fedora 26 (ncurses 6.0) where the 6.1 version fails on the chtype check, but the 6.0 version is fine (unsigned int).

This pertains to the line in extconf.rb:

 convertible_int('chtype', [["#undef MOUSE_MOVED\n"]]+curses) or abort

Strangely the definition for chtype hasn't changed:

#if 1 && defined(_LP64)
typedef unsigned chtype;
typedef unsigned mmask_t;
#else
typedef uint32_t chtype;
typedef uint32_t mmask_t;
#endif

It's not clear why convertible_int can't detect the type.

I just tested this example:

https://github.com/ruby/curses/blob/master/sample/form.rb

https://github.com/ruby/curses/blob/master/sample/form.rb

However had it crashed for me:

001_form.rb:46:in `driver': Unknown command (Curses::UnknownCommandError)
from 001_form.rb:46:in <main>

(I renamed it locally to 001_form.rb - makes it easier for me to test the examples
when I number them.)

The line was:

else
   form.driver(ch)

I added some debug statement. I could cause this error by hitting the "Ende" (End)
key.

The key value for this was:

360

Is this the regular behaviour of curses to crash like that?

To me it is a bit strange that form.driver(ch) would not work.

I did not test it with other keys but I assume that some other keys also cause
this issue.

Edit: Actually I tested it with page down too. That also led to the same problem,
hopefully it can be reproduced by someone else.

Hi,

I am currently the maintainer of the ncursesw gem (https://rubygems.org/gems/ncursesw, originally based on the seemingly unmaintained: http://ncurses-ruby.berlios.de/) which we host as part of the sup-heliotrope project: https://github.com/sup-heliotrope/ncursesw-ruby.

It seems that this project is overlapping, but I think that our implementation is slightly more extensive. Would it be worth the effort to join the projects?

Complement issue: sup-heliotrope/ncursesw-ruby#22

Regards, Gaute

This article describes 256 colours for ncurses:

https://reversed.top/2019-02-05/more-than-256-curses-color-pairs/

The third and last piece is that most of curses examples you’ll find are using old API and there
seems to be no (at least easy to find) documentation on how to transition to the new one.

There are applications which use new API, it’s just documentation that didn’t really caught up.

Could the README mention whether this is supported or possible in the ruby-ncurses wrapper?

It seems like that subpad's begin_x and begin_y coordinates aren't relative to the parent pad but relative to the terminal screen if parent pad has been put on the screen using refresh. Not sure if it's the intended behavior. This example raises exception (failed to create window):

pad = Curses::Pad.new(2, 2)
pad.refresh(0, 0, 1, 1, 1, 1)
sub = pad.subpad(1, 1, 0, 0)

And this one works & looks the same on the screen - subpad is at (1, 1) in the terminal window and at (0, 0) relatively from parent pad.

pad = Curses::Pad.new(2, 2)
pad.refresh(0, 0, 1, 1, 1, 1)
sub = pad.subpad(1, 1, 1, 1)

This one works too and looks the same (refresh after pad & subpad creation):

pad = Curses::Pad.new(2, 2)
sub = pad.subpad(1, 1, 0, 0)
pad.refresh(0, 0, 1, 1, 1, 1)

It makes it difficult to create new subpads if parent pad is currently displayed on the screen, as you must know where it resides, making new subpad behave similarly to subwin. You can't put new subpad using relative coordinates to parent pad. Maybe I'm using it in a wrong way?

Hi

I can use 8 colors provided, but I would very much like to use more than that. Is there any nice way for me to define my own colors? I can't find any reasonable resources on this idea.
This might be helpful: can_change_colors?, returns false for me, but I have no idea why.

puts "\e[1;90mt\e[0m"

will render me a nice, grey "t"

Gem document is incomplete and lack of examples.