Skip to content

Conversation

piechologist
Copy link

Added parameter q=1 to the draw requests. The terminal won't respond with an escape sequence that we don't read anyway.

The Kitty graphics protocol requires the terminal to send a confirmation in form of an escape sequence (see here). KittyTerminalImages.jl does not capture the response so, something like _Gi=1234;OK\ may leak into the following Julia prompt in some terminals (I'm currently beta-testing Ghostty). The quiet mode solves this issue.

Added parameter `q=1` to the draw requests. The terminal won't respond with an escape sequence that we don't read anyway.
@simonschoelly
Copy link
Owner

Thanks- I wanted to test this myself in kitty - but I cannot see the response sequence that kitty sends back - do you know where that is actually written to? I tried out read(stdin) after plotting something but it does not look like its written to Julia's input stream.

@piechologist
Copy link
Author

Yeah, not all terminals do it the same way. For instance, Ghostty prints the response but WezTerm does not. AFAIK, some terminals don't even send a response at all.

You can put the terminal into raw mode. I think the best way is the following code from TerminalExtensions.jl:

using REPL: Terminals
function foo()
    term = Terminals.TTYTerminal("xterm", stdin, stdout, stderr)
    Terminals.raw!(term, true)
    Base.start_reading(stdin)
    write_kitty_image_escape_sequence(stdout, # ... your request here
    data = readuntil(stdin, "\e\\") # read the response up to ST (the string terminator)
    return data
end

BTW, thank you for this extension! I use SixelTerm.jl with WezTerm but I think the Kitty protocol is the way to go.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants