01:28:03  * AtumTquit (Remote host closed the connection)
01:59:54  <Bakkot>bterlson: yes, I think so. at least, without leaking, though not necessarily with the same asymptotic performance.
02:27:22  * not-an-aardvarkquit (Quit: Connection closed for inactivity)
03:27:48  * jmdyckquit (Remote host closed the connection)
04:42:14  * not-an-aardvarkjoined
05:17:03  * gskachkovquit (Quit: gskachkov)
05:44:47  <Domenic>I thought there were things you could do with weak map that you couldn't do with weak ref (and obviously vice versa)
05:45:42  <ljharb>yeah you'd need the combo of both to do EVERYTHING :cackle:
05:55:58  <Bakkot>I think that depend on whether you are defining "things" to include "building data structures with particular asympotic time performance" or "building data structures which do not require any manual 'GC' ". with just WeakRefs you can simulate WeakMaps, except that a.) they will be linear access rather than log, and b.) they will require you to notice that manually GC values when the WeakRef you're using as a key stops referring.
05:55:59  <Bakkot>(which, given executors, you could do without a seperate thread.)
05:56:35  <Bakkot>anyway, my impression was still that engines are pretty strongly against weakrefs, because of not wanting to expose gc semantics to user code.
07:06:06  * gskachkovjoined
09:01:19  * abernixquit (Ping timeout: 248 seconds)
09:22:44  * abernixjoined
10:25:11  * mylesborinsquit (Quit: farewell for now)
10:25:42  * mylesborinsjoined
10:27:22  * not-an-aardvarkquit (Quit: Connection closed for inactivity)
11:32:56  * jmdyckjoined
12:54:39  * gskachkovquit (Quit: gskachkov)
12:55:16  * gskachkovjoined
13:36:39  * gskachkovquit (Quit: gskachkov)
13:41:39  * isHavvyjoined
13:41:45  * Havvyquit (Ping timeout: 240 seconds)
13:42:02  * gskachkovjoined
14:01:53  * gibson042joined
14:02:26  * AtumTjoined
14:32:08  * bradleymeckjoined
14:43:43  * gskachkovquit (Quit: gskachkov)
15:45:15  * pandemquit (Remote host closed the connection)
16:17:20  * gskachkovjoined
16:32:18  <TabAtkins>I've said a few times now that we need the ability to intercept gets/sets/deletes on Maps and Sets. It would have made WebIDL capable of using real Maps and Sets, rather than just look-alikes.
16:34:36  * bradleymeckquit (Quit: bradleymeck)
16:42:37  * bradleymeckjoined
16:57:20  * pandemjoined
16:57:31  * gskachkovquit (Quit: gskachkov)
17:13:03  * gskachkovjoined
19:00:47  * abernixquit (Quit: Sleepin')
19:10:08  * abernixjoined
19:41:59  * bradleymeckquit (Quit: bradleymeck)
19:51:32  * bradleymeckjoined
22:19:46  * Fishrock123joined
22:33:30  * bradleymeckquit (Quit: bradleymeck)
23:08:46  * Fishrock123quit (Remote host closed the connection)