Fix source location for JSXEmptyExpression nodes (fixes #248)

[jlongster]

Q	A
Bug fix?	yes
Breaking change?	no
New feature?	no
Deprecations?	no
Spec compliancy?	no
Tests added/pass?	no
Fixed tickets	#248
License	MIT

Turned out this is an easy fix. Looks like it wasn't appropriately using this.state as it should. See #248.

[hzoo]

👋 @jlongster - can you also add a fixture test for this?

[jlongster]

Sure, need to learn how to add a test first so will take me a bit.

[DrewML]

@jlongster All of the tests in Babylon just have fixtures that match a source string to the serialized AST. Basic steps are:

1. Add a new directory with a descriptive name under one of the subfolders in test/fixtures/jsx
2. Add an actual.js file in that folder with the code to use as a test case
3. Run npm run test-only, which will automatically generate an expected.json for you
4. Inspect the AST output to expected.json to verify it looks correct

It also looks like you'll need to regenerate the results for test/fixtures/jsx/basic/10, since they're currently committed without the proper location info. You can delete that expected.json and re-run npm run test-only to regenerate.

[hzoo]

We can add this steps in CONTRIBUTING like I did in Babel via https://github.com/babel/babel/blob/master/CONTRIBUTING.md#writing-tests

[jlongster]

@hzoo The current tests seem to be failing:

% npm run test-only

> [email protected] test-only /Users/james/projects/babylon
> ava


  2343 passed
  1 failed

  jsx/basic/10
  Error: /Users/james/projects/babylon/test/fixtures/jsx/basic/10/actual.js: [
  {
    "type": "CommentBlock",
    "value": " this is a comment ",
    "start": 4,
    "end": 27,
    "loc": {
      "start": {
        "line": 1,
        "column": 4
      },
      "end": {
        "line": 1,
        "column": 27
      }
    }
  }
] !== null (leadingComments/expression/0/children/expression/0/body/program)
    runTest (test/utils/runFixtureTests.js:62:13)
    Test.fn (test/utils/runFixtureTests.js:14:22)

This is on the latest master without my changes. I see this failure in my CI too. I suppose I can fix it along with my PR, but is that expected? I don't see that failure in other PRs, but I don't understand why I can reproduce it by checking out master without my changes.

[DrewML]

I don't understand why I can reproduce it by checking out master without my changes.

@jlongster It's not super well documented (or at all?), but the tests right now actually run against the artifact spit out by Rollup. We have some clean up to do there.

For the time being, try running npm run build && npm run test-only on master, and things should be passing.

Regarding the current tests failing in CI, haven't looked deep into it yet, but my suspicion is that your fix addressed an issue that we didn't notice when that fixture was last checked in. If you verify the AST in jsx/basic/10 is correct (and that it was wrong before) you can delete the expected.json and run npm run test-only to generate the new fixture to include in your PR.

Edit: Yeah, the fixture checked in had some issues. Empty object for loc, and no start or end prop on the JSXEmptyExpression node. Just regenerate that fixture and you should be set.

Thanks for contributing, btw!

[jlongster]

That makes sense, I figured it was something like that! No worries, keeping docs up-to-date and making these kinds of things takes a huge amount of time so I don't really mind figuring some things out.

[jlongster]

Done!

[danez]

Nice catch, thanks.

[jlongster]

Any chance this could be merged?

[hzoo]

👍 thanks @jlongster