Let's take Java as a point of reference, since you brought it up. Imagine the JDK were opinionated about which authors were allowed to write packages that use FFI. And people wanting to use it would be instructed to build a relationship to said authors to get their code blessed to use FFI. It is wholly inconceivable. I'm sure you'd see quite a few "Leaving Java" posts over it.
Here's what Oracle says[0]: "Technically, nothing prevents your program from calling into sun.* by name." And I don't think I ever accessed sun.* when writing Java. So the example seems both ill-suited and far-fetched to me.
One of the first things I did in Elm was writing a tiny JS snippet to get a missing piece from the browser. It was something I needed and my site wouldn't have worked without it. I later switched to the Elm implementation once that became available.
Now given the stated goals of the Elm project, I will be unable to repeat this. Which means I expect I'd be stuck if I again wanted to access the browser API before the Elm people got around to implement that part. And that wholly changes my view of the project.
I'm using Elm for a toy project. If I'm blocked, I go do something else. What a pity though.
