What happens when an implementation-defined limit is reached?

[annevk]

Typically this is not defined, in part because editors are not always aware limits might be in play and in part because they do not want to constrain implementation choices. But there is something to be said that if implementations choose the same strategy, they should behave in the same way.

E.g., if implementations decide to throw an exception if an input is larger than X bytes, perhaps the exception should at least be the same, even if X is not.

There is some precedent for this: RangeError exception for ArrayBuffer object allocation. But at the same time JavaScript doesn't deal with any of the other out-of-memory situations, thereby allowing for exceptions, crashing the process, no-opping, or some other behavior.

I'm not sure what is best here. For out-of-memory a process crash seems quite clean, but we're not at a process-per-origin yet so it's not that clean.