This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The structured cloning algorithm isn't totally clear on what should happen when an element is modified after the algorithm in e.g. 6.3.1 invoked a getter. I think the best solution would be to iterate over all key/values first and then invoking the structured cloning algorithm on that list. Consider this example: map = new Map() map.set("b," {get b() { map.remove("a"); }); map.set("a", "b");
Just to clarify here: 1. The structured cloning algorithm may call getters and proxy traps and who knows what else, so the order of operations it specifies is observable. 2. The List and Record types in the ES spec are mutable, and in particular operations on a Map can modify its internal [[MapData]] List and assign to the [[key]] and [[value]] fields of the Records in that List. Is the intended interaction of language like "For each [...] that is an element of source, run the following substeps" with mutation explained somewhere? (Alternatively, is it supposed to be obvious? In that case I apologize for being such a dunce) Of course objects are mutable too so step 8 also seems vague.
How it works in Firefox: - for Objects, we make a copy of the list of own enumerable properties at the time step 8 starts executing - then, just before step 8.2, we do a [[Has]] check, and if the property is gone, we skip 8.2-8.4 for that property. That's already implemented. Here's what we're implementing for Map and Set: - for Map, take a snapshot of the keys and values in the [[MapData]] at the time step 6 starts executing. Then we never consult the Map object or its internal data again. In step 6.3 we write exactly the keys and values in the snapshot, regardless of what has been done to the Map. - same goes for Set: shallow copy on entry to step 7.
Shouldn't this just be specified in the JS spec?
Adam, Jason, Tom, can I convince any of you to make a PR against https://github.com/whatwg/html? Map/Set have already been fixed by Adam. See https://github.com/whatwg/html/commit/367707af5cd76864be9224e4b5a769fbf44fed96.
(In reply to Anne from comment #4) > Adam, Jason, Tom, can I convince any of you to make a PR against > https://github.com/whatwg/html? > > Map/Set have already been fixed by Adam. See > https://github.com/whatwg/html/commit/ > 367707af5cd76864be9224e4b5a769fbf44fed96. Reading Jason's comment above I realize that my fix didn't go far enough (or with enough specificity) so someone should fix that too (basically I forgot that the spec modifies the records in place).
(In reply to Adam Klein from comment #5) > (In reply to Anne from comment #4) > > Adam, Jason, Tom, can I convince any of you to make a PR against > > https://github.com/whatwg/html? > > > > Map/Set have already been fixed by Adam. See > > https://github.com/whatwg/html/commit/ > > 367707af5cd76864be9224e4b5a769fbf44fed96. > > Reading Jason's comment above I realize that my fix didn't go far enough (or > with enough specificity) so someone should fix that too (basically I forgot > that the spec modifies the records in place). Handled this problem in https://github.com/whatwg/html/pull/82 (though I can't sign up to fixing the rest of this today)
I think this should be all clear now.