Permit %%placeholder%% in left-hand-side of a let declaration

[Zalathar]

Q	A
Fixed Issues?	Fixes #12717
Patch: Bug Fix?	Yes
Major: Breaking Change?
Minor: New Feature?
Tests Added + Pass?	Yes
Documentation PR Link
Any Dependency Changes?
License	MIT

This change adds support for template syntactic placeholders in the left-hand-side of a let declaration (e.g. let %%lhs%% = 7).

This already works for var and const, but wasn't working for let because the special handling needed to disambiguate let-as-keyword vs let-as-identifier didn't know to treat a following placeholder as though it were a variable.

[babel-bot]

Build successful! You can test your changes in the REPL here: https://babeljs.io/repl/build/39237/

[codesandbox-ci]

This pull request is automatically built and testable in CodeSandbox.

To see build info of the built libraries, click here or the icon next to each commit SHA.

Latest deployment of this branch, based on commit c54793b:

Sandbox	Source
babel-repl-custom-plugin	Configuration
babel-plugin-multi-config	Configuration

[Zalathar]

There are a few different ways to implement this.

I didn't like the idea of overriding isLet in the placeholder plugin, because then I'd have to copy-paste a bunch of code from the base implementation, which seems needlessly messy.

So my choice was between adding adding a very narrow hook method that the plugin could override, or just adding ad-hoc code directly in the parser itself.

I went with the ad-hoc code because it's very straightforward and compact, and that seems worth the downside of having placeholder code that lives outside the placeholder plugin.

[Zalathar]

I'd like to add some explicit test cases for let % x and let %%x%% when the placeholders plugin is not being used, but I'm not quite sure where to put them.

[nicolo-ribaudo]

I would really prefer to keep this code in placeholders.js, even if we would then have to duplicate this.isContextual("let").

This isn't an hot code path (usually the placeholders plugin is not enabled), and even when perf matters the isContextual function is not expensive.

Btw, thanks for also adding tests for the let %%LHS%%; case!

[Zalathar]

Btw, thanks for also adding tests for the let %%LHS%%; case!

Just to be clear, I have only added tests for let %%LHS%% = %%RHS%%; (with initializer), for the initializer forms with var/const, and for the negative case let %LHS; where the % should be parsed as modulo.

I had also written tests for let %%LHS%%; (without initializer), but I didn't add them in this PR because they fail, with or without this patch! That's a separate issue that will need a separate fix.

[nicolo-ribaudo]

and for the negative case let %LHS; where the % should be parsed as modulo

Sorry I was tired 🙃

[Zalathar]

I'm rather baffled by my new failure test (let context 3) producing a subtly different error under CI.

[nicolo-ribaudo]

I think it could be because GitHub runs CI not on your branch, but on the result of merging your branch in main.
We recently merged 8cf0a75, which could cause the test failure.
Could you try rebasing on main?

[nicolo-ribaudo]

Thanks!