kitten icat: Failed to open controlling terminal

[iambumblehead]

Is there a way to use kitten icat inside an ncurses application?

Using kitten icat inside an ncurses application here gives the error Failed to open controlling terminal

kitten icat --align left "$img_path"

Per this thread I've also tried the following commands which result in different errors,

kitten @ --to=$KITTY_LISTEN_ON icat --align left "$img_path"
# Error: icat is not a known subcommand for @. Use --help to get a list of valid subcommands.
kitten icat --align left "$img_path" @ --to=$KITTY_LISTEN_ON
# Error: Failed to open controlling terminal with error: open /dev/tty: device not configured

thanks in advance for any help or advice

[kovidgoyal]

On Wed, Mar 27, 2024 at 08:18:32PM -0700, iambumblehead wrote: Is there a way to use kitten icat inside an ncurses application?

Yes, just run it, and make sure your application is not doing any IO with /dev/tty while running it.

Using `kitten icat` inside an ncurses application here gives the error `Failed to open controlling terminal` ```bash kitten icat --align left "$img_path";; ```

That's not an ncurses application, thats probably nvim, which works in client server mode and has no access to the tty. If you want to display images in nvim use images.nvim

Per [this thread](#2426 (comment)) I've also tried the following commands which result in different errors, ```bash kitten @ --to=$KITTY_LISTEN_ON icat --align left "$img_path";; # Error: icat is not a known subcommand for @. Use --help to get a list of valid subcommands. kitten icat --align left "$img_path" @ --to=$KITTY_LISTEN_ON ;; # Error: Failed to open controlling terminal with error: open /dev/tty: device not configured ```

Remote control cant be used to display images, see https://sw.kovidgoyal.net/kitty/remote-control/

[iambumblehead]

the parent application is vifm and basically this same issue is discussed at stackoverflow here https://stackoverflow.com/questions/76024179/issue-with-previewing-images-in-vifm-on-wayland-with-kitty-on-arch-linux

[kovidgoyal]

On Wed, Mar 27, 2024 at 08:29:00PM -0700, iambumblehead wrote: the parent application is vifm and basically this same issue is discussed at stackoverflow here https://stackoverflow.com/questions/76024179/issue-with-previewing-images-in-vifm-on-wayland-with-kitty-on-arch-linux

Then you need to get vifm to allow you to run processes with access to the terminal device. Or get it to suppor thte kitty image protocol natively.

[kovidgoyal]

On Wed, Mar 27, 2024 at 08:29:00PM -0700, iambumblehead wrote: the parent application is vifm and basically this same issue is discussed at stackoverflow here https://stackoverflow.com/questions/76024179/issue-with-previewing-images-in-vifm-on-wayland-with-kitty-on-arch-linux

Oh and that stack overflow answer is incomplete. You dont need a terminal if you can pass the window dimensions manually, as demonstrated by zsh -c 'setsid kitten icat --use-window-size 100,30,3000,2000 --place ***@***.*** --transfer-mode=file logo/kitty.png' You will of course need to replace the various dimensions and the filename. But as I said in my previous reply, its easier just to use the tty.

[kovidgoyal]

And see 399a9d6

[iambumblehead]

Thanks for replying and helping me. The approach suggested at the vifm issue linked from stackoverflow is working here. 'Will try setting window-size and removing /dev/tty redirections next,

kitten icat \
  --place "${img_wh}@${img_tl}" \
  --align left \
  --transfer-mode=stream "$img_path" >/dev/tty </dev/tty

[iambumblehead]

after updating kitty; --use-window-size is working without errors

[iambumblehead]

removing the /dev/tty redirections (either one or both) results in an error

kitten icat \
  --use-window-size "${kicat_wh_cells/x/,},${kicat_wh_pixels/x/,}" \
  --place "${img_wh}@${img_tl}" \
  --align left \
  --transfer-mode=stream "$img_path"

error

Error: The --place option can only be used with a single image, not 2

[kovidgoyal]

On Wed, Mar 27, 2024 at 10:03:16PM -0700, iambumblehead wrote: removing the /dev/tty redirections (either one or both) results in an error ```bash kitten icat \ --use-window-size "${kicat_wh_cells/x/,},${kicat_wh_pixels/x/,}" \ --place "${img_wh}@${img_tl}" \ --align left \ --transfer-mode=stream "$img_path" ``` error ```bash Error: The --place option can only be used with a single image, not 2 ```

Add --stdin=no

[iambumblehead]

Add --stdin=no

error is not seen after adding that, however image is no longer rendered --instead long text fills the screen

[iambumblehead]

updating the command to use --transfer-mode=file but no change occurred

[kovidgoyal]

On Wed, Mar 27, 2024 at 10:16:59PM -0700, iambumblehead wrote: > Add --stdin=no error is not seen after adding that, however image is no longer rendered --instead long text fills the screen

that will be because vifm is mangling the escape codes it gets from icat before passing them to the tty. So stick to using the %N mode with /dev/tty.

[iambumblehead]

It is agreeble if the discussion is closed as "resolved" however one point of mention; the same script renders sixel image format from inside vifm plus other terminal emulators without /dev/tty redirections It turns out the sixel function below also displays noise when used with vifm+iterm2

image_to_sixel_magick () {
    img_path=$1
    img_wh=$3

    export MAGICK_OCL_DEVICE=true
    if [[ -n "$is_cmd_magick" ]]; then
        magick \
            -background "rgba(0,0,0,0)" \
            "$img_path" \
            -geometry "$img_wh" \
            sixel:-
        echo ""
    elif [[ -n "$is_cmd_convert" ]]; then
        convert \
            -channel rgba \
            -background "rgba(0,0,0,0)" \
            -geometry "$img_wh" \
            "$img_path" \
            sixel:-
        echo ""
    else
        fail "$msg_cmd_not_found_magickany"
    fi
}

[kovidgoyal]

Then you should ask the vifm developers to pass through kitty graphics protoco escape codes since they do that for sixel escape codes already.

[iambumblehead]

yes, kitty is working correctly here and the issue originates at vifm.

This issue occurs when vifm is configured to provide both top and left position to the image viewer script. I assumed this was a bug and reported it to vifm.

The developer explained that, when vifm is configured to provide both top and left position (which the script uses to define --use-window-size "${kicat_wh_cells/x/,},${kicat_wh_pixels/x/,}") it makes assumptions about the returned value and escapes it.

There is no issue at kitty and the related vifm discussion is here vifm/vifm#731 (comment)

Thank you for facilitating and helping me :)