windows: add support for native virtual terminal

[perillo]

Unfortunately I'm unable to test that, when ENABLE_VIRTUAL_TERMINAL_PROCESSING is available, the Windows console supports all the escape codes used by the compiler and standard library.

From https://web.archive.org/web/20230000000000*/https://learn.microsoft.com/en-us/windows/console/console-virtual-terminal-sequences, the official documentation was only added in 2022.

[matu3ba]

This reads like either 1. a version check should be used here or 2. an assertion should be added.

[perillo]

The idea was to advertise this feature, but probably it is not necessary.

A version check is not necessary, since if SetConsoleMode returns 0, then the feature is not supported.

[matu3ba]

1. Does this affect the use case of using dumb as argument to the windows console?
2. Did you compare how other applications are doing it, ie Windows Terminal? It sounds very fishy, that 2 kernel32 calls are necessary to query the console state.

[perillo]

1. Does this affect the use case of using dumb as argument to the windows console?

I'm not aware of any dumb arguments for the windows console. Note that I use Windows only for testing software compatibility.

2. Did you compare how other applications are doing it, ie [Windows Terminal](https://github.com/microsoft/terminal)? It sounds very fishy, that 2 kernel32 calls are necessary to query the console state.

AFAIK, Windows Terminal enables this the feature by default.

The problem, that I did not consider, is how to detect Windows Terminal.
With the current PR, escape codes will be disabled (not tested) when a Zig program runs on a Windows Terminal, since calling GetConsoleMode will fail (not tested).

See microsoft/terminal#1040.

Is anyone using Zig on a Windows Terminal?

Thanks.

[perillo]

1. Does this affect the use case of using dumb as argument to the windows console?

   2. Did you compare how other applications are doing it, ie Windows Terminal? It sounds very fishy, that 2 kernel32 calls are necessary to query the console state.

Note that the first call is to get the current state and the second it to set it. It is the same as with Unix termios.

Also, about the Console API and Windows Terminal: calling GetConsoleMode should not fail.

[squeek502]

Windows Terminal seems to set ENABLE_VIRTUAL_TERMINAL_PROCESSING automatically.

In regular cmd.exe:

In Windows Terminal:

---

Test program:

const std = @import("std");

pub extern "kernel32" fn SetConsoleMode(in_hConsoleHandle: std.os.windows.HANDLE, in_Mode: std.os.windows.DWORD) callconv(std.os.windows.WINAPI) std.os.windows.BOOL;

const ENABLE_VIRTUAL_TERMINAL_PROCESSING = 0x0004;

pub fn main() !void {
    const handle = std.io.getStdErr().handle;
    var mode: std.os.windows.DWORD = undefined;
    if (std.os.windows.kernel32.GetConsoleMode(handle, &mode) == 0) {
        @panic("couldn't get console mode");
    }

    const enabled = mode & ENABLE_VIRTUAL_TERMINAL_PROCESSING != 0;
    std.debug.print("{}\n", .{enabled});
    std.debug.print("\x1b[32;1myep\x1b[0m\n", .{});

    mode |= ENABLE_VIRTUAL_TERMINAL_PROCESSING;
    const should_now_be_enabled = SetConsoleMode(handle, mode) != 0;

    std.debug.print("{}\n", .{should_now_be_enabled});
    std.debug.print("\x1b[32;1myep\x1b[0m\n", .{});
}

[perillo]

@squeek502 There is some documentation in https://ss64.com/nt/syntax-ansi.html.

[squeek502]

All I was showing was that Zig setting ENABLE_VIRTUAL_TERMINAL_PROCESSING is unnecessary in Windows Terminal--it's already set. The escape sequences are valid/correct and are there to show how the console renders them.

To make a more general comment about the PR, I'm unsure if setting ENABLE_VIRTUAL_TERMINAL_PROCESSING is something Zig should be doing unconditionally like this. This is similar to #7600 (see also #12400 and #14411) in that it affects all processes run in the console session, not just the current process, so it's not clear cut if attempting to force ENABLE_VIRTUAL_TERMINAL_PROCESSING to be enabled is always the 'right' thing to do. If we can get away with just #15282, it seems like that might be the 'safer' way to go.

[expikr]

To make a more general comment about the PR, I'm unsure if setting ENABLE_VIRTUAL_TERMINAL_PROCESSING is something Zig should be doing unconditionally like this. This is similar to #7600 (see also #12400 and #14411) in that it affects all processes run in the console session, not just the current process, so it's not clear cut if attempting to force ENABLE_VIRTUAL_TERMINAL_PROCESSING to be enabled is always the 'right' thing to do. If we can get away with just #15282, it seems like that might be the 'safer' way to go.

conversely, this would also mean that if utf8 is becoming unconditionally prescribed, enabling virtual terminal in the same vein would also be fair game?

[squeek502]

This is similar to #7600 (see also #12400 and #14411) in that it affects all processes run in the console session, not just the current process

I was wrong about this, apologies for not checking my assumptions. See #20172 for a revivial of the spirit of this PR.