A Better Way to Code
Mike Bostock

Awesome, looks like a great candidate for bringing javascript to the big data/machine learning domain!

I wonder if this could be easily/efficiently integrated with big number libraries, to get around the floating point limitations of JS.

I also wonder how this can integrate with React (or similar); seems like simply rendering to a DOM element in a block would suffice, but perhaps JSX could become part of the d3.express syntax. There is some impedance mismatch because you are using two-way binding and React uses bound values; but that can be wrapped.

For example, given a<Range/> input component that accepts value and onChange(newValue) as attributes (amongst others), perhaps typing <Range value={foo}/> automatically translates to rendering <Range value={foo.currentValue} onChange={n => foo.dispatchValue(n)}/> (I have no idea how you implemented the DAG nodes).

Really happy to see this in any case, it looks like a great improvement over the (never released) demo that Bret Victor showed so long ago. Can’t wait to play with it!

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.