Windows terminal: thin box characters appear as question marks

[timgilbert]

Running the sample file via bb -Sdeps '{:deps {io.github.paintparty/bling {:mvn/version "0.4.2"}}}' -e "(require '[bling.sample]) (println (bling.sample/sample))" in Windows Terminal 1.21.32331.0, I see some extended characters rendered as question marks:

[...]

It looks like the thinner boxes from the demo are the ones affected. I don't know much about Windows character sets but am happy to test any fixes to this.

[paintparty]

Hi @timgilbert - thank you for reporting this and your willingness to test fixes.

Would you mind telling me the name of the font you are using with Windows Terminal? Or better yet try to zip the actual font file and attach it to a comment in this thread. I'm curious to see if I can replicate the question mark rendering on my side (mac).

Also, could you try using this font in Windows Terminal?
JetBrainsMono-2.304.zip

I'm wondering if those box drawing chars are outside the specific IBM PC / WGL4 subset that is mentioned here

If that is the case, the fix might be to use the slightly thinner versions of those (in the Bling lib) that would be in the subset.

[paintparty]

@timgilbert - also wondering if this setting specific to Windows Terminal mentioned here would help?

[timgilbert]

Sorry for the delay! The font I'm using in Windows Terminal is Cascadia Mono, which is the default font. I've tried all the other ones in the "Appearance / Font face" preference and they all show question marks. I tried installing the JetBrains Mono fonts from the zip file above, but I still see the question marks when I set the terminal to use them.

I've tried toggling the "builtin glyphs" setting on and off, but it has no impact. (I tried with the JSON file as in the above link, then noticed there's just a Preferences / Appearance setting for it; neither had any effect.)

[timgilbert]

I do see evidence that Windows Terminal supports these characters, eg microsoft/terminal#14654. Interestingly, when I copy and paste the box characters from github into my terminal, eg from this line, they show up just fine in the terminal:

I've also tried running this with clj instead of bb and got the same result, so it doesn't seem to be a babashka-specific thing.

[timgilbert]

I have noticed one odd discrepancy between clj and bb, though. The Clojure REPL prints box characters just fine, but bb prints them as question marks:

But running clj -Sdeps '{:deps {io.github.paintparty/bling {:mvn/version "0.4.2"}}}' -e "(require '[bling.sample]) (println (bling.sample/sample))" still prints the question marks.

[paintparty]

Thank you for taking the time to test this!

From a cursory glance at the slack thread the borkdude referenced, it seems unlikely that there is a bb-specific problem.

Just to clarify, do you get the question marks in Windows Terminal when you do the following from a repl in your terminal (requires clojure 1.12.0)? The bling samples should be printed after you call (bling.sample/sample):

➜  ~ clj

Clojure 1.12.0

user=> (add-libs '{io.github.paintparty/bling {:mvn/version "0.4.2"}})
[io.github.paintparty/bling]

user=> (require 'bling.sample)
nil

user=> (bling.sample/sample)

[paintparty]

Also, looking at this page, it says:

Windows Terminal is a modern host application for the command-line shells you already love, like Command Prompt, PowerShell, and bash (via Windows Subsystem for Linux (WSL)).

My knowledge of windows is limited, but I'm curious if Windows Terminal allows you to choose between using Command Prompt or PowerShell, and if that would be a factor in the box-drawing character rendering. It sounded like the user in the slack thread referenced above was using cmd.exe(Command Prompt), and they were not seeing Korean chars render properly.

[timgilbert]

Hmm, well I'm not sure what I did, but I've suddenly got the proper characters to display in Clojure, but not babashka:

In case anyone comes across this in the future, the relevant bit from my settings.json is:

            {
                "guid": "{574e775e-4f2a-5b96-ac1e-a2962a402336}",
                "hidden": false,
                "name": "PowerShell",
                "showMarksOnScrollbar": true,
                "source": "Windows.Terminal.PowershellCore"
            },

Possibly a Clojure upgrade had something to do with this? In any event, bb still shows the question marks.

I tried the same sequence in cmd.exe and Windows PowerShell (v5.1.x, versus "regular" PowerShell which is 7.4.6) and the command shell had no effect.

Seems like the remaining bit of the problem is in some difference between how babashka and clj process extended characters on Windows, versus anything specific to bling. Thanks for your help! I think this ticket can be closed; I'll open a babashka ticket and link it here.

[paintparty]

Thank you @timgilbert