Experiment with lebab (transpiling *to* ES6)

[Fil]

As an experiment (prompted by #179), I applied lebab to d3-geo, which resulted in the branch es6 in https://github.com/Fil/d3-geo/commits/es6

I've used a private fork because I don't think we want this for d3-geo.

Fun results though: the build loses 2.1% & minified build loses 2.9%.

ES5:

• 88441 d3-geo.js
• 35626 d3-geo.min.js

ES6:

• 86528 d3-geo.js
• 34590 d3-geo.min.js

log

for rule in arrow arrow-return for-of for-each arg-rest arg-spread obj-method obj-shorthand no-strict let; do
  lebab --replace 'src/**/*.js' --transform $rule
  yarn test && git commit -a -m"lebab --replace 'src/**/*.js' --transform $rule"
done

patch for the rule let
ignore the rule class

for rule in commonjs template ; do
  lebab --replace 'src/**/*.js' --transform $rule
  yarn test && git commit -a -m"lebab --replace 'src/**/*.js' --transform $rule"
done

rule default-param needs a fix for src/clip/rectangle.js
ignore rule destruct-param (same issue as rule default-param)

for rule in  includes ; do
  lebab --replace 'src/**/*.js' --transform $rule
  yarn test && git commit -a -m"lebab --replace 'src/**/*.js' --transform $rule"
done

[mbostock]

I am strongly opposed to shipping transpiled-to-ES5 code. This will not be reliable, and not something that I want to spend effort supporting.

[gordonwoodhull]

@Fil transpiled d3-geo to ES6!

[mbostock]

Oh, I totally misread, sorry!

[Fil]

(no need to keep this open)