For some specific sizes it scales to 1.5 line heights

hmm, I thought I'd fixed that

@pxeger @graffe ato.pxeger.com/… is more correct

(Sorry I didn't realize it auto scaled and just thought it was stuck, but ofc it scales since I've used the python version too and there you normally need multiple lines)

I did indeed fix that, but it wasn't included in v0.1.16 so it's not been deployed

Hyped for 0.1.17

I can deploy it without needing to release v0.1.17

1.17 has to be delayed, of course

Why?

minecraft joke

Oh ok

I would really like to copy snippets with proper class for syntax highlighting

I have already deployed that

Oh nice

@mousetail ok, done now

you may need to hard-refresh (ctrl+shift+R)

TY

Unfortunatly it seems like language highlighting still doesn't work. In fact any html tag with a class attribute is simply ignored on SE

hmm

I'll have to change the way the code formatting is generated then

Is there any specific reason you chose to generate html tags instead of standard markdown?

because HTML is more consistent than markdown

Not when run through a sanitizer

if someone wrote code containing three backticks, it would break markdown

Only when the backticks appear on their own line

yes

You can use the 4-spaces version then, I don't think it can be ambiguous

maybe if you have multiple blank lines in a row

that doesn't work for leading and trailing blank lines I think

TIO has fairly complicated logic for this already, which I could just copy/reimplement

Ok fair enough every method sucks

github.com/TryItOnline/tryitonline/blob/master/usr/share/…

That except with a extra case for trailing blank lines

I could probably submit a PR for that

I'd prefer backticks where possible rather than indented code

and they can handle leading/trailing blank lines, which means <pre><code> is less often needed

I can do that. Check if there is a line with backticks, then generate backticks, else if there are no leading or trailing double line breaks use spaces, else use <pre><code>

This could be a challenge on main lol

I think certain unprintable characters would also require <pre><code> to be used

True

Might require some iteration to find which those are

tab is the one I just tested, I suspect most others get ignored by SE anyway

and tab is definitely the most important one to get right

TIO has a list, I can probably just copy it

basically all it probably needs to do is check for /^```$/m or /\t/

Is there a way to test the frontend without building the entire app?

npm run dev

inside the frontend directory

and create a file called .env.development.local with the content NEXT_PUBLIC_ATO_BASE_URL="https://ato.pxeger.com"

@pxeger In frontend or in the root?

in frontend

ok thanks it works

@pxeger github.com/attempt-this-online/attempt-this-online/pull/77

It seems the <!-- language: lang-xxx --> comments work on <pre><code> formats too.

Oh that's a lot simpler lol

I still like the readability benefits of using the indent and backtick formats though

yeah, I'm just adding to your PR

:thumbsup:

← 60 messages moved from The Nineteenth Byte

@pxeger I see you are rewriting the backend in rust. As you know I love rust, but it seems like a lot of efort. What is the reason behind that decision?

I started under the mistaken belief that Rust would be able to both run an asynchronous web server and still be fork-safe (which Go is not)

now it turns out Rust can't do that safely either (actually, no language can if it uses more than one thread), but I've come up with a workaround

and now the only reason I'm carrying on is that I'm enjoying it

it may never get merged

Yea rust has great thread safety but fork safety is really hard to guarentee

Realistically, my program would only do a little bit of stuff between fork and exec, but I haven't thought through if any of that could still cause unsafety

What type of stuff do you need to do?

@pxeger (I mean - do that, all in one executable. My workaround was just to use a new executable that doesn't do anything after fork)

@mousetail set up a container

I really haven't enjoyed having my entire container logic written in Zsh

things like creating cgroups, namespaces, then entering them

I understand how that would be simple enough that you think you should be able to write it safely but still complex enough that you can't really trust any implementation to actually be safe

yup

@Deadcode Can you write a container manager in .NET regex for me? lol

Is regex fork safe? Guess we'll find out

@pxeger lol

How much disk space does ATO take up total?

[root@atovps ~]# df -h /
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda1        19G   15G  3.2G  82% /

some of that will be the base OS, not ATO, though

about 11GB is the language images

Ah :-) interesting.

I wonder how much is taken up on TIO by the program-run-result cache.

the rest is about 10MB