00:22 <jfhbrook> I've never had that problem

00:23 <jfhbrook> ljharb: suppose you had a tree that you were representing with instances of classes, and you wanted to be able to serialize that thing with JSON.stringify

00:23 <jfhbrook> ljharb: would you rather have a toJSON method that explicitly filters out properties you don't want to serialize? or would you make those properties non-enumerable?

03:11 <ljharb> probably a combination of both.

03:11 <ljharb> in both cases you can test it, but an explicit toJSON is probably easier to test.

03:11 <ljharb> and also instead of a blacklist, you'd use a whitelist.

03:26 <jfhbrook> in the case of a blacklist, and even more so for a whitelist, does it bother you that now there's a big list of properties kicking around somewhere, when you could just inspect the object to get the lists of (enumerable) properties yourself?

03:26 <jfhbrook> in the case of using non-enumerable properties, does it bother you that this effects not just serialization but cloning as well?

03:27 <jfhbrook> (I made the argument that it was silly to expect a clone operation to pick up non-enumerable properties since it doesn't pick up prototypes either

03:27 <jfhbrook> )

03:28 <jfhbrook> (but maybe you want to keep those concerns separated?)

03:30 <ljharb> no, it doesn't bother me

03:30 <ljharb> using reflection means that at runtime, i can change the behavior of the method without actually overwriting the method.

03:30 <ljharb> now, that said, i'd write tests that use reflection to verify that the property list is correct.

03:30 <ljharb> as for a "clone" - there's no concept in JS of cloning yet.

03:31 <ljharb> but, i'd think it would have to grab all own property descriptors (enumerable or not), as well as the [[Prototype]].

03:32 <jfhbrook> indeed ljharb which is why I think that expectation is so silly---all you can really do is Object.assign

03:32 <jfhbrook> and yeah ljharb there's example code that does something *like* that on mdn

03:32 <ljharb> wait, which expectation

03:32 <ljharb> you can do `Object.assign` right now

03:33 <ljharb> but you can now, thanks to me, do `Object.create(Object.getPrototypeOf(obj), Object.getOwnPropertyDescriptors(obj))`

03:33 <ljharb> so it's not that silly.

03:34 <jfhbrook> oh that's cool

03:34 <jfhbrook> that's in es5? 6?

03:34 <ljharb> ES5, ES5, and ES8/ES2017 respectively

03:34 <jfhbrook> oh, getOwnPropertyDescriptors is new

03:35 <jfhbrook> yeah I see

03:35 <ljharb> yup

03:35 <ljharb> https://www.npmjs.com/package/object.getownpropertydescriptors

03:35 <ljharb> obv i have the polyfill for it :-p

03:35 <jfhbrook> can you use this without polyfilling?

03:36 <ljharb> oh sure. it doesn't shim by default.

03:36 <jfhbrook> oh word

03:36 <ljharb> you have to call `.shim()` to do that.

03:36 <jfhbrook> yeah no I see it now

03:36 <jfhbrook> cool

03:36 <ljharb> i even have an API contract for that :-) https://github.com/es-shims/es-shim-api

03:36 <jfhbrook> nerrrrrrd

03:36 <jfhbrook> XD

03:36 <jfhbrook> very nice though

03:37 <ljharb> true story

03:37 <ljharb> and ty

03:37 <jfhbrook> are people following this? like have you gained traction?

03:38 <jfhbrook> man I hate github's new font

03:40 <alva> looks like it's trying to use the system font

03:50 <ljharb> jfhbrook: which, es-shim-api? only one or two people have taken a real interest so far. but i'm building up to the point where it will be enough of a critical mass that it doesn't need much.

03:53 <jfhbrook> nice

20:40 <pikajude> ljharb: we figured it out

20:40 <pikajude> the fake redis library we were using adds between 1000 and 2000ms of random delay to every call

20:40 <pikajude> that obviously doesn't show up in profiling output

20:41 <jfhbrook> lolwut

20:41 <purr> lolololol

20:41 <jfhbrook> kinda hilarious that it ended up being something so pedestrian pikajude

20:43 <pikajude> kinda hilarious

21:01 <pikajude> jfhbrook: I've wasted probably 6 hours waiting for this testsuite