supports-color breaks coloring

[NaNraptor]

This line over here defines an ANSI extended color set for bright color which doesn't display correctly: (by incorrectly I mean it's just bold white color)

https://github.com/visionmedia/debug/blob/5c7c61dc0df0db4eb5de25707d8cd1b9be1add4f/src/node.js#L169

It shouldn't(?) have a traling ;1 after ${colorCode}. So the line should like:

const prefix = ` ${colorCode}m${name} \u001B[0m`;

This fixes the bug (for me :) )

EDIT: This supports my observation 256 colors

Also from the article:

Note that the bright versions of the background colors do not change the background, but rather make the foreground text brighter. This is unintuitive but that's just the way it works.

and the ;1 option only applies to \u001b[44 ie background colors, while the \u001b[38 option for foreground colors does not have such thing

[Qix-]

• 1m with 38;...m works just fine, in theory.
• which emulator are you using?
• screenshot please.

[NaNraptor]

Sorry for the late reply, here are some screenshots.

Here I'm using debug v4.1.1 installed with npm install debug.
The result is the same when using git bash(with cmd console) and powershell:

What I found strange is that the +NNNms has the correct color, while the namespace didn't, I narrowed it down to that line I pointed to in my initial comment.

Below I'm using my fork as the version installed with npm install https://github.com/NaNraptor/debug
Here is the result when I use the version from my fork:

[Qix-]

Could you do me a favor and change this line:

https://github.com/visionmedia/debug/blob/5c7c61dc0df0db4eb5de25707d8cd1b9be1add4f/src/node.js#L168

To this:

-const colorCode = '\u001B[3' + (c < 8 ? c : '8;5;' + c);
+const colorCode = '\\e[3' + (c < 8 ? c : '8;5;' + c);

Re-run and post the results here?

[NaNraptor]

Yes of course.

First run is when ;1 is not in the code, the second run is with it:

[Qix-]

Those look fine, they should be working. This is a bug in CMD.exe.

I'll open a ticket with Microsoft later.

[NaNraptor]

You are in fact correct, I tested it on a mac and it works just fine

[Qix-]

Bold has always been quite broken on Windows. Here's an open issue that probably has to do with this bug.

microsoft/terminal#109

For posterity, here's a bug where bold-off (21m) was completely missing that I filed in 2017.

microsoft/WSL#2174

So, it's an ongoing process.

---

cc @zadjii-msft is this a new problem that I should file, or is this the same as microsoft/terminal#109?

[zadjii-msft]

So, microsoft/terminal#109 is mostly just tracking the rendering of bolded text, which is still somewhere down the backlog and unsupported.

However, the bug here is that combining 256-color values (38;5;210m for ex) and boldness (1m) doesn't work correctly. I don't know when for a fact that was fixed, but I do believe it's fixed in Insider's now, and will be generally available with the next Windows update after 1809:

[Qix-]

Wonderful, thanks for the quick response as always @zadjii-msft :)

I'm going to close this as fixed. Thanks for the report @NaNraptor, good find.

[webuniverseio]

@qix, what if I make a PR that removes boldness based on environment variable? I'm working in a corporate environment and very likely that windows updates that will bring this feature will take a while :(. And it will depend on each of my coworkers to update their machine 😢. Thanks

[Qix-]

@webuniverseio I wouldn't approve that, unfortunately. I don't want to introduce more environment variables when we're working to remove them.

You can, however, do FORCE_COLOR=0 as an environment variable