Fatal: Micro could not initialize a Screen.

[tajsnowberry]

Description of the problem or steps to reproduce

Fatal: Micro could not initialize a Screen.

Install from curl https://getmic.ro | bash and run it.

Specifications

Running KDE Plasma on Arch Linux, I am using Konsole
Commit hash: b976385
OS: Arch Linux
Terminal: Konsole

[rilysh]

I can't able to reproduce your problem, seems works fine for me, no issue at all. Can you try micro on xfce terminal, by doing pacman -S xfce4-termimal and see if it works on there or not. Though I'm not in Arch, currently on Debian with KDE and using micro on Konsole, everything looks good for me.

[tajsnowberry]

It works on Budgie

[fhdk]

I can reproduce on Arch

• Arch Linux KDE (Plasma 5.25.4 Framework 5.97.0) -> X11 or Wayland -> Konsole

  charater set not supported
  Fatal: Micro could not initialize a Screen
  

• Arch Linux KDE (Plasma 5.25.4 Framework 5.97.0) -> X11 or Wayland -> Xfce terminal

  charater set not supported
  Fatal: Micro could not initialize a Screen
  

---

No issues with

• Manjaro KDE (Plasma 5.25.4 Framework 5.97.0) -> X11 -> Konsole

[Brawdunoir]

Hi, I am having same issue on Arch Linux Sway (Wayland) using Alacritty terminal.

$ micro --version
Version: 2.0.11
Commit hash: 225927b9
Compiled on August 09, 2022

Installed using pacman but same issue using your installation method @tajsnowberry

Any news on this issue ?

[vazub]

Same issue on OpenBSD -current (either installed from packages, or built from scratch from upstream master):

obsd$ micro
function not implemented
Fatal: Micro could not initialize a Screen.

obsd$ micro -version
Version: 2.0.13
Commit hash: 68d88b5
Compiled on Unknown

[rosco-pc]

Same issue on OpenBSD -current (either installed from packages, or built from scratch from upstream master):

obsd$ micro function not implemented Fatal: Micro could not initialize a Screen.

obsd$ micro -version Version: 2.0.13 Commit hash: 68d88b5 Compiled on Unknown

For OpenBSD is fails due to : golang/go#63900

[4a6f656c]

@vazub @rosco-pc - the most likely cause is code that is calling syscall.Syscall* - either directly in this code base, in a dependency or an out-of-date dependency. Running under egdb with a breakpoint on the syscall.Syscall* functions should identify the source of the call.

[rosco-pc]

On Mon, 29 Apr 2024 07:51:29 -0700 Joel Sing ***@***.***> wrote:

@vazub @rosco-pc - the most likely cause is code that is calling `syscall.Syscall*` - either directly in this code base, in a dependency or an out-of-date dependency. Running under `egdb` with a breakpoint on the `syscall.Syscall*` functions should identify the source of the call.

It is running syscall.Syscall6 in the tcell dependency (also from zyedidia :)) From a mail chain on openbsd ports:

…

> On Sun, 7 Apr 2024 12:21:14 -0400 > George Koehler ***@***.***> wrote: > > > On Sun, 7 Apr 2024 15:57:26 +0200 > > Rob Schmersel ***@***.***> wrote: > > > > > I tracked it down to the use of syscall in the tcell package > > > file tscreen_bsd.go (line 46) > > > > > > > > > > fd = uintptr(t.out.(*os.File).Fd()) > > > > if _, _, e1 := syscall.Syscall6(syscall.SYS_IOCTL, fd, > > > > ioc, tios, 0, 0, 0); e1 != 0 { e = e1 > > > > goto failed > > > > } > > > > Joel Sing patched our lang/go to allow SYS_IOCTL in Syscall, but > > not in Syscall6. I tried changing Syscall6 to Syscall in > > tscreen_bsd.go. The "function not implemented" error stopped > > appearing, but > > > I think there is a reason why Syscall6 was used instead of Syscall > (i.e. number of arguments). Just removing the last 3 arguments now > makes the last argument (tios) be a pointer (at least according to > the syscall package description) which I'm not sure is the > intention and probably the reason for the black screen :) > > So I guess I need to look at the syscall package to figure what > changes where made for Syscall. Will give it go later today. > > > - micro got stuck on a black screen; I needed to close the > > terminal. > Got the same after patching > > > - I can't add my patch to the port, because "make update-patches" > > doesn't see it. The file is outside WRKSRC, in > > ***@***.*** > > > > Yeah, the tcell package is a dependency that gets pulled in during > compilation. We need to request upstream to patch or get tcell as an > own package once it is figured out what is the cause (it definitely > used to work before :)) > > > --- tscreen_bsd.go.orig.port Sun Apr 7 11:53:03 2024 > > +++ tscreen_bsd.go Sun Apr 7 11:57:57 2024 > > @@ -43,7 +43,7 @@ > > tios = uintptr(unsafe.Pointer(t.tiosp)) > > ioc = uintptr(syscall.TIOCGETA) > > fd = uintptr(t.out.(*os.File).Fd()) > > - if _, _, e1 := syscall.Syscall6(syscall.SYS_IOCTL, fd, > > ioc, tios, 0, 0, 0); e1 != 0 { > > + if _, _, e1 := syscall.Syscall(syscall.SYS_IOCTL, fd, > > ioc, tios); e1 != 0 { e = e1 > > goto failed > > } > > @@ -61,7 +61,7 @@ > > tios = uintptr(unsafe.Pointer(&newtios)) > > > > ioc = uintptr(syscall.TIOCSETA) > > - if _, _, e1 := syscall.Syscall6(syscall.SYS_IOCTL, fd, > > ioc, tios, 0, 0, 0); e1 != 0 { > > + if _, _, e1 := syscall.Syscall(syscall.SYS_IOCTL, fd, > > ioc, tios); e1 != 0 { e = e1 > > goto failed > > } > > @@ -94,7 +94,7 @@ > > fd := uintptr(t.out.(*os.File).Fd()) > > ioc := uintptr(syscall.TIOCSETAF) > > tios := uintptr(unsafe.Pointer(t.tiosp)) > > - syscall.Syscall6(syscall.SYS_IOCTL, fd, ioc, > > tios, 0, 0, 0) > > + syscall.Syscall(syscall.SYS_IOCTL, fd, ioc, tios) > > t.out.(*os.File).Close() > > } > > if t.in != nil { > > @@ -108,8 +108,8 @@ > > dim := [4]uint16{} > > dimp := uintptr(unsafe.Pointer(&dim)) > > ioc := uintptr(syscall.TIOCGWINSZ) > > - if _, _, err := syscall.Syscall6(syscall.SYS_IOCTL, > > - fd, ioc, dimp, 0, 0, 0); err != 0 { > > + if _, _, err := syscall.Syscall(syscall.SYS_IOCTL, > > + fd, ioc, dimp); err != 0 { > > return -1, -1, err > > } > > return int(dim[1]), int(dim[0]), nil > > >

[anoduck]

Same issue on OpenBSD -current (either installed from packages, or built from scratch from upstream master):
obsd$ micro function not implemented Fatal: Micro could not initialize a Screen.
obsd$ micro -version Version: 2.0.13 Commit hash: 68d88b5 Compiled on Unknown

For OpenBSD is fails due to : golang/go#63900

Been running into this issue for several years, finally I now know why, and I am grateful.

[niten94]

The issue on OpenBSD is reported in #3557 and different with the problem reported here, but Micro should be able to run on OpenBSD by the next release.

In the meantime, I've also sent a message to one mailing list including a partial way to patch 2.0.14, so that someone may patch the version of Micro provided in OpenBSD ports:
https://marc.info/?l=openbsd-ports&m=176137734514024&w=2

[niten94]

To the Linux users on this thread that had issues: does Micro 2.0.11 or newer still crash like below on any system?

charater set not supported
Fatal: Micro could not initialize a Screen

If it does:

• Please post the 1st line if it's different with above.
• Otherwise, please paste this command in bash and post the output: printf '"%s"\n' "$LC_ALL" "$LC_CTYPE" "$LANG"

The installation method isn't relevant if the screen can't be initialized, but the OS and DE are important to know.

Micro shouldn't abort if the 3 variables don't contain a dot. Otherwise, it still should run if charset is utf8, utf-8, or uppercase, as specified in .charset[@modifier] at the end of the variables.

[dimk90]

I had this problem on ArchLinux. The cause was an extra space in $LANG.

Running printf '"%s"\n' "$LC_ALL" "$LC_CTYPE" "$LANG" showed:

""
""
"en_US.UTF-8 "

After changing "en_US.UTF-8 " to "en_US.UTF-8" the problem was resolved.

---

micro: 2.0.15-dev
ArchLinux (6.18.2)
No DE ($TERM: 'linux')