sourcemaps and runtime variables misaligned, yielding difficult debugging

[cdaringe]

Can you reproduce the problem with latest npm?

yes

Can you still reproduce it?

yes

Description

The sourcemaps appear to work, however, the browser runtime (chrome, in my case) does not honor variable names.

^ Notice how parseUrl is defined and valid, yet I cannot access it in the render scope.

^ in fact, _url.parse is a thing. i don't understand why url was mangled to _url or why a new var parseUrl was not put into scope.

I realize that this might be a browser issue, vs a create-react-app issue, but can't say for certain because it may related to CRA's source-map generation.

Expected behavior

Have vars accessible in scope as written in source

Actual behavior

vars are mangled and hidden from the user.

Environment

Run these commands in the project folder and fill in their results:

1. npm ls react-scripts (if you haven’t ejected): 0.9.x
2. node -v: 7.4.1
3. npm -v: 4.12.1, yarn 0.19.1

Then, specify:

1. Operating system: el cap
2. Browser and version: chrome 56.0.2

Reproducible Demo

• clone https://github.com/cdaringe/react-scripts-source-map-wonky-isms
• cd
• yarn
• yarn start
• open dev tools. it should already break at the correct location to see the failure mode

[gaearon]

Yes, unfortunately it's a known issue with sourcemaps in Babel (and everywhere, really), and one of the reasons I argued against including sourcemaps. However, a lot of people find them very valuable so we opted to include them despite the drawbacks.

I think there was some sort of change to sourcemap spec to retain variable mapping information. But I don't know if either Babel, Webpack, or Chrome are taking advantage of it.

Probably relevant issues:

• Chrome: https://bugs.chromium.org/p/chromium/issues/detail?id=327092 (seems like it landed?)
• Babel: (no idea, help find it?)
• Webpack: (no idea, help find it?)

cc @sokra @hzoo who might know more.

[gaearon]

As for your particular case I'm not even sure if support for mangled mapping will help.

^ in fact, _url.parse is a thing. i don't understand why url was mangled to _url or why a new var parseUrl was not put into scope.

It is necessary because of ES6 spec. In ES6, imports are “live” so the only way to implement it in ES5 is to have a variable indirection for the import itself. It’s the same reason you see abc.default when you import abc from 'abc': Babel needs to put the import behind an object so that it stays “live” per spec.

See also http://www.2ality.com/2015/07/es6-module-exports.html (search for “live” on the page).

[hzoo]

I only remember babel/babel#3658 which broke webpack and also referenced before in #1188

[Timer]

As I recall, the patch to webpack (source-list-map) simply disregarded the new mappings.
Maybe we can find someone willing to add support?

[gaearon]

It would be nice to verify that without Webpack in the middle, sourcemaps produced by Babel give meaningful results in Chrome. Just so we have a confirmation it's worth doing the work on Webpack side.

[Timer]

I think we just need to test with full-blown sourcemaps enabled (during development, I may be wrong tho), because source-list-map is what is used for one-mapping-per-line iirc. webpack might also strip this elsewhere.

[Timer]

@cdaringe do you have time to test if sourcemaps produced by Babel (without webpack) function correctly as @gaearon described above? Help would be greatly appreciated.

[cdaringe]

i made an example here, only to realize that of course pure babel is ok--its babel-loader w/ uglify that warrant these difficult source maps. i also did an example with babili. all is green 🍏, but it's a biased test because babel & babili are maintaining my symbol names.

[cdaringe]

nm babili is mangling some stuff & its busted per the same failure mode.

pure babili example above:

check out that wild local var!

[gaearon]

For now I don't think we care about what happens with Babili as we don't ship with it.
Also I am more interested in mapping working in DEV mode (in which we don't even Uglify).

[cdaringe]

then the first example should suffice :). however, theres no bundler or uglification/mangling w/ pure babel, so no reason for the failure mode described above to affect it.

[gaearon]

What is the next actionable step here? Do we need to file an issue with one of the projects? Which one?

[cdaringe]

sorry, my remarks didn't add clarity. lets fix that! your original thoughts were ~~accurate.

• babel, without ES2015 module imports, works fine with sourcemaps.
• babel, with ES2015 module imports, does not work w/ sourcemaps

therefore there should be a followup in babel-land on this. this can be easily demonstrated by cloning & running this tiny example. because we wanted a webpack-free example, i stub in a window.require that simply returns React. the example uses a pure babel build (using CRA's babel config) w/ a fake require. Again, Chrome cannot show in-scope imports, such as Component.