JSX parsing panics when using the spread operator

[kitsonk]

Describe the bug

Using the spread operator in a JSX block causes a panic (at least in Deno).

Input code

const A = () => {
  return <div>{...[]}</div>;
};

Expected behavior

The code to be emitted without a panic and something like:

"use strict";
const A = () => {
    return React.createElement("div", null, []);
};

Actual behaviour

❯ RUST_BACKTRACE=1 deno bundle --no-check spread.tsx
Bundle file:///Users/kitsonk/github/deploy_dg/spread.tsx
thread 'main' panicked at 'cannot access a scoped thread local variable without calling `set` first', /Users/kitsonk/.cargo/registry/src/github.com-1ecc6299db9ec823/scoped-tls-1.0.0/src/lib.rs:168:9
stack backtrace:
   0: std::panicking::begin_panic
   1: swc_ecma_transforms_react::jsx::Jsx<C>::jsx_elem_child_to_expr
   2: core::iter::traits::iterator::Iterator::find_map::check::{{closure}}
   3: <core::iter::adapters::chain::Chain<A,B> as core::iter::traits::iterator::Iterator>::next
   4: <alloc::vec::Vec<T,A> as alloc::vec::spec_extend::SpecExtend<T,I>>::spec_extend
   5: swc_ecma_transforms_react::jsx::Jsx<C>::jsx_elem_to_expr
   6: <swc_ecma_transforms_react::jsx::Jsx<C> as swc_ecma_visit::VisitMut>::visit_mut_expr
   7: swc_ecma_visit::visit_mut_stmts
   8: <swc_ecma_transforms_react::jsx::Jsx<C> as swc_ecma_visit::VisitMut>::visit_mut_expr
   9: swc_ecma_visit::visit_mut_var_declarators
  10: <swc_visit::Optional<V> as swc_ecma_visit::Fold>::fold_module
  11: swc_bundler::bundler::load::<impl swc_bundler::bundler::Bundler<L,R>>::load_transformed
  12: <core::iter::adapters::map::Map<I,F> as core::iter::traits::iterator::Iterator>::next
  13: deno::module_graph::Graph::emit_bundle
  14: deno::bundle_module_graph
  15: <core::future::from_generator::GenFuture<T> as core::future::future::Future>::poll
  16: <core::future::from_generator::GenFuture<T> as core::future::future::Future>::poll
  17: <core::future::from_generator::GenFuture<T> as core::future::future::Future>::poll
  18: deno::main

Version

swc_common = "0.11.4"
swc_ecmascript = "0.46.0"

Additional context

[kdy1]

Seems like swc_ecma_utils::HANDLER is not configured?

[kdy1]

Btw, it does not look like an erroneous input.

[NamPNQ]

I think the expected behavior is to raise an error

For information, babel it will raise errors:
Spread children are not supported in React.

[kitsonk]

@NamPNQ TypeScript transpiles it without error. Also there is a big difference between an error and a panic.

[kdy1]

@kitsonk
It panics because swc_ecma_utils::HANDLER is not configured

[kitsonk]

Ok, I will look into that. It seems we don't have that set across several projects.

The input isn't an error though, as I mentioned, TypeScript transforms it without error.

[kdy1]

swc also transpiles it without an error.
I'm not sure if it's related to HANDLER.

[kdy1]

Can you try configuring swc_ecma_utils::HANDLER?

[kitsonk]

I am trying to, but we have never used HANDLER.set() anywhere in Deno before. I think I have figured out what I need to do to create a handler, but I am not exactly sure where it needs to be scoped in our code.

Specifically when we bundle, we first create a swc_bundler::Bundler and then take the .bundle() and run it through a swc_ecmascript::codegen::Emitter. Where does the scoping of the HANDLER.set() need to occur?

[kdy1]

I think scope should be the invokation ofjsx transform.
I'm not sure if it will fix the issue. I asked trying it because I can't guess a cause of panic.

[dsherret]

It was because swc_ecma_utils::HANDLER wasn't set. This API seems very hidden and I'm not sure ideal (ex. typescript's api returns emit diagnostics as an array on a property of the emit result). I implemented it in Deno in denoland/deno#12773 by collecting the diagnostics and then returning an error after transforming if a diagnostic with a certain level appeared. Works ok, I think, now that we know about it.

The error we get now is: Spread children are not supported in React.. I see Babel gives that error, but as Kitson pointed out TypeScript seems to handle this transform. I don't think there's any harm in implementing it.

[kdy1]

I'll do so, as I prefer behavior of tsc of it of babel.

[binoche9]

How difficult would the fix for this be (if someone with no context of the swc codebase would try to fix this)?

[kdy1]

It's not hard, but not sure how much it will be for others

[swc-bot]

This closed issue has been automatically locked because it had no new activity for a month. If you are running into a similar issue, please create a new issue with the steps to reproduce. Thank you.