* Pass `dirname` as extra metadata to preset constructor.
Sometimes a preset would like to know where it should resolve relative paths from (e.g. https://github.com/tleunen/babel-plugin-module-resolver) and this extra information makes that possible.
* Test for `dirname` passed into preset constructor
This adds a check for `dirname`’s existence and correctness to the
`resolve-addons-relative-to-file` test, and serves as a minimal example
of a path-aware preset.
When you write
```
for (const x of l) {
setTimeout(() => x);
}
```
we need to add a closure because the variable is meant to be block-scoped and recreated each time the block runs. We do this.
However, we also add the closure when no loop is present. This isn't necessary, because if no loop is present then each piece of code runs at most once. I changed the transform to only add a closure if a variable is referenced from within a loop.
This change ensures that when hoisting an identifier we do not hoist it up to
the first no shadowed function, but rather up to the bound shadowed function
* Flip default parameter template
YMMV, I saved ~10b on a 2kb library. Not noticeable at the small scale, by why not do it anyway?
I've (unscientifically) found that flipping the default parameter conditional yields better gzip results. I think this is due to the slightly longer string it can now repeatedly match:
```js
// old
var param = arguments.length <= 0 || void 0 === arguments[0] ? null : arguments[0]
--------------------------------------------------------------^
// new
var param = arguments.length > 1 && void 0 !== arguments[1] ? arguments[1] : null
------------------------------------------------------------------------^
```
Though it's entirely likely gzip will also choose up to the index of the arguments if you many default parameters at different indexes.
* Update tests
* Removed unnecessary 'return' statements.
Returning a 'Promise' value in 'promise.then()' makes promise chains.
Used memory and promises are not garbage collected
until finishing 'helpers.asyncToGenerator'.
* Update test
* Fix class inheritance in IE9 & IE10 (T3041)
Internet Explorer 9&10 do not support __proto__ at all, don't have
Object.setPrototypeOf(), but have Object.getPrototypeOf().
Because of this setting the prototype is not possible, which makes the
babelHelpers.inherits() function to set __proto__ although not supported.
Afterwards Object.getPrototypeOf() is used, but this one is not
respecting the "custom" property __proto__ that we set.
The solution is to check for __proto__ first and afterwards fallback to
Object.getPrototypeOf().
* Do the same logic in babel-helper-replace-supers
* Fix tests
* Extract creation of prototype nodes to small helper
I previously tried an approach to scope bindings from var to scope but
it didn't catch all cases. This is evident in this bug:
https://phabricator.babeljs.io/T2892
Where even after transforming a const to a var we still get an error
that it's read-only.
This approach will go through and delete every existing let and const
binding and creates a new one with the kind "var"
Rather than repeating the original location when we exit a node, we need to restore the previous parent’s original line location, since the source map format denotes the start location.
