Jot tests fail on JavaScript

[lpil]

I have not investigated yet but seeing as it's pure Gleam it means there's some bug in the compiler or the standard library causing it here.

https://github.com/lpil/jot

[BradLewis]

So I had a look into this, and it looks like this is due to the difference in behaviour between the string.split_right implementation between erlang and javascript. The test(s) in question use unicode spaces (U+00A0), and the erlang implementation will keep the space, whereas the javascript version will strip it.

The relevant code is in the following function:

// jot.gleam
fn inlines_to_html(html: String, inlines: List(Inline), refs: Refs) -> String {
  case inlines {
    [] -> html
    [inline, ..rest] -> {
      html
      |> inline_to_html(inline, refs)
      |> inlines_to_html(rest, refs)
      |> string.trim_right
    }
  }
}

If we look at the bad test, with javascript we go from

"<p><em> a "
to
"<p><em> a"

whereas in erlang we keep the space.

I'm not sure what the desired behaviour is here though in order to fix it.

Erlang REPL:

❯ erl
Erlang/OTP 26 [erts-14.2.5] [source] [64-bit] [smp:10:10] [ds:10:10:10] [async-threads:1] [jit] [dtrace]

Eshell V14.2.5 (press Ctrl+G to abort, type help(). for help)
1> string:trim("<p><em> a ", trailing).
"<p><em> a "

Node REPL:

❯ node
Welcome to Node.js v18.12.1.
Type ".help" for more information.
> "<p><em> a ".trimRight()
'<p><em> a'

[lpil]

Good detective work! That is a rather annoying difference. 🤔

[BradLewis]

Yeah definitely not ideal. I'm not sure how/if we would want to resolve this difference. Is there precedence elsewhere in the standard library where there is/was different behaviour between JS and Erlang?

[lpil]

We want them to behave the same here.

[BradLewis]

In that case I have created a PR to rework the javascript behaviour to match the erlang behaviour.