Good question!

Two reasons:

documentation.js uses Babel to get an abstract syntax tree, which is what it uses to run inference - we look at the AST structure and derive parameters and so on. Babel transforms-away a lot of the things we'd want to document - destructuring parameters, for instance, are transformed into a single parameter and then destructuring within the function. For inference to work, we need to have a look at the original source, not a transformed version.
When we link to github, run lint, or output an error, we rely on correct line numbers - line numbers that match the original source code, so that you can find the error. Babel's transformation changes the code and the line numbers.

We're using babelify as kind of a hack: we use module-deps to find dependencies, and to find dependencies in es6 code, it needs to invoke babelify - otherwise, module-deps would choke on es6 syntax and import statements and so on. So we aren't actually using babelify for its usual purpose of transforming code for output - we're only transforming code to figure out where dependencies are.

We parse code with babylon in lib/parsers/javascript, which gives us the AST, and that's the source of truth for inference.

why "un-transform babelify transformed source"? #357

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions