A thought re sub refs passed in

[shadowcat-mst]

While I think most of the warnings would/should still apply, I think you could get -mostly- timely destruction for them.

The trick would be to keep a WeakMap or WeakSet of the javascript functions proxying to the subref on the JS side, plus a hash or Hash::Util::FieldHash of the subrefs themselves on the perl side, and to periodically check to see if the WeakXXX entry has gone away and release the subref if so.

Sadly I don't immediately see a way to convince QuickJS to do a full GC, so you'll be dependent on its decision for that, but since it's a hybrid refcounting + mark&sweep GC (and perl programmers tend to reflexively -not- create circular references if we can avoid it ;) there seems like a decent chance that the function object will go away at a sensibly early moment - or at least often enough that at least with a little care such a feature would be worth turning on for quite a lot of users.

The question of whether this is worth the additional complexity is left as an exercise to the maintainer - but nonetheless, I'm pretty sure it would -work- at least.

(I am not currently using this module but was having a look around; it -is- now filed in my brain as the first thing to try when I need something in this space, but that doesn't mean you should listen to me if your actual users don't care ;)

(I have a project in mind that absolutely would end up using this, hence looking around, but whether the yak ordering will result in me writing it any time soon is unknown)

[FGasper]

Of potential interest:

bellard/quickjs#337

It's an interesting idea. Probably not something I'd do myself--to many other irons in the fire--but I'd be happy to look at a PR.

Alternatively, as I no longer use Perl for $work, someone with a bigger stake in this module could assume maintainership.