Remove `transformOptions`

[gnestor]

Partially fixes https://github.com/n-riesco/jp-babel/issues/2

Theoretically, if options are not passed to babel.transform then babel should use the .babelrc options. I haven't fully tested to determine if this in fact is happening. I know that module import is not working in the notebook within a project that has a qualified .babelrc...

[n-riesco]

@gnestor Thank you for investigating this. Unfortunately, this isn't working in my machine.

I've also tried:

var transformOptions = {
    babelrc: true,
};

but it doesn't work either.

A solution to #3 would be that jp-babel looks for .babelrc and sets transformOptions accordingly if it finds one.

[gnestor]

@n-riesco I just updated the PR to use jp-babel's babel-preset-es2015 by providing an absolute path to it. I've confirmed that I no longer get Error: Couldn't find preset "es2015".

[gnestor]

Regarding https://github.com/n-riesco/jp-babel/issues/3, I started working on finding and importing config from .babelrc. One question: How do we get the CWD of the jupyter notebook? Given that (say cwd), we could:

function findFile(file, dir) {
    var result = fs.readdirSync(dir).find(result => result === file);
    if (result) return path.resolve(result);
    return findFile(file, path.join(dir, '..'));
}

function getBabelConfig(path) {
  if (!path) return null;
  try {
    return JSON.parse(fs.readFileSync(path, 'utf-8'));
  } catch(err) {
    return null;
  }
}

var babelrc = getBabelConfig(findFile('.babelrc', cwd));

var transformOptions = babelrc || {
    presets: [path.resolve(path.join(__dirname, "..", "node_modules", "babel-preset-es2015"))],
};

This traverses up to the directory tree (starting from the notebook CWD) looking for a .babelrc and uses its config if it exists. Otherwise, it defaults to jp-babel's babel-preset-es2015.

[n-riesco]

The problem with the notebook CWD is that it only makes sense for local kernels (remote kernels, in principle, cannot access notebook paths).

I understand the need for making the use case of local kernels as convenient as possible. But this requires help from the frontend.

jp-babel, ijavascript and jp-coffeescript provide a flag to set the session working directory:

jp-babel --jp-working-dir=path

When this flag isn't provided, the session working directory is set to the current working directory. This means that the following example will launch the jupyter notebook with jp-babel kernels that use /path/to/my/project as the session working directory.

$ cd /path/to/my/project
$ jp-babel

---

Sorry I'm digressing. The point I wanted to make is that I think we should use:

var babelrc = getBabelConfig(findFile('.babelrc', process.cwd()));

and let the frontend set the kernel working directory. If the frontend doesn't help, the user has the workaround of setting this manually with jp-babel --jp-working-dir=path.

[gnestor]

@n-riesco Thanks, very helpful. I just updated the PR again. Now, if I run notebook in a directory with a .babelrc file (anywhere up its directory tree), jp-babel will use that config and if not, it will default to using babel-preset-es2015. I think this fixes https://github.com/n-riesco/jp-babel/issues/2 and https://github.com/n-riesco/jp-babel/issues/3. Give it a try!

[gnestor]

@n-riesco Updated! Here's a test notebook (should've done that from the start!). https://gist.github.com/gnestor/5ee40cf0009e5aa037948f7c4ad6044c

[n-riesco]

@gnestor Hi! I've just tested the PR and it works for me.

Before I merge it, please, could you update the AUTHORS file? I ask contributors to do so, to confirm their consent to distributing their contributions under the LICENSE terms.

Thanks for fixing issues #2 and #3.

[gnestor]

@n-riesco Thank you for starting this project and helping me with these issues.