使用模块捆绑与完整 D3 捆绑时 D3.v4 力图中断

Question

I've beed fiddling around with D3 force-directed graphs, eg https://bl.ocks.org/mbostock/4062045 in D3.v4. All works fine with the newest complete D3 build/bundle.

I want to reduce the size of JS libraries loaded, and thus ended up using the rollup-based https://www.npmjs.com/package/d3-v4-bundler . All my other, unrelated D3.v4 graphs work fine with rollup-bundled "hand-picked" modules and methods, but something goes wrong with d3-force.

Already early on in the code when attempting to use d3-force methods:

var simulation = d3.forceSimulation()
  .force("link", d3.forceLink().id(function (d) { return d.id; }))
  .force("charge", d3.forceManyBody())
  .force("center", d3.forceCenter(svgwidth / 2, svgheight / 2));

An error is thrown: "Uncaught TypeError: Cannot read property 'force' of undefined".

I've tried pretty much every combination of modules and methods in d3-v4-bundler, but only if i export everything from "d3", ie use the whole thing, the force-directed graph works. Even bundling all modules separately doesn't work. "d3-force" itself, to my knowledge, requires "d3-collection", "d3-dispatch", "d3-quadtree" and "d3-timer" but these are all correctly bundled via d3-v4-bundler.

Is there possibly some other (external?) vital dependency to "d3-force" that could go missing in the bundling process that causes d3.forceSimulation().force to break? Or could this be a problem rooted in d3-v4-bundler or rollup itself?

Answer 1

Found an answer to my problem, but not to very the reason behind it.

For some reason, either some contents of only d3-force or the combined contents in my selection of d3 modules get minified with Uglifyjs in d3-v4-bundler in a manner that breaks the "force" decleration. When removing the "-c negate_iife=false" parameters from d3-v4-bundler's uglifyjs command, even the minified code works for graphs generated with d3-force. Of course the minimization rate is not as good, but obviously something in value consolidation breaks otherwise.

Not a big expert myself, but would be useful if the actual part that breaks could be tracked down. "-c negate_iife=false" worked fine for other module bundles until d3-force was added, or at least so it seemed to me.