00:19:23  * bradleymeckquit (Quit: bradleymeck)
00:20:38  * Guest59quit (Quit: My Mac has gone to sleep. ZZZzzz…)
00:25:56  * plutoniixjoined
00:29:17  * Guest59joined
00:32:37  * Guest59quit (Client Quit)
00:35:00  * Guest59joined
01:07:59  * Guest59quit (Quit: My Mac has gone to sleep. ZZZzzz…)
01:08:10  * unixpicklequit (Quit: My Mac has gone to sleep. ZZZzzz…)
01:08:37  * unixpicklejoined
01:09:18  * unixpicklequit (Client Quit)
01:10:05  * Guest59joined
01:31:42  * joyeejoined
01:44:33  * wadeyquit (Quit: ZNC - http://znc.in)
01:45:03  * wadeyjoined
02:03:49  * unixpicklejoined
02:42:06  * joyeequit (Remote host closed the connection)
02:42:48  * joyeejoined
02:42:50  * joyeequit (Remote host closed the connection)
02:46:25  * unixpicklequit (Quit: My Mac has gone to sleep. ZZZzzz…)
02:50:48  * joyeejoined
04:58:54  * plutoniixquit (Remote host closed the connection)
05:01:03  * bradleymeckjoined
05:03:34  * bradleymeckquit (Client Quit)
05:34:46  * unixpicklejoined
05:36:29  * unixpicklequit (Client Quit)
06:14:23  * joyeequit (Remote host closed the connection)
06:15:54  * joyeejoined
06:15:56  * joyeequit (Remote host closed the connection)
06:43:15  * joyeejoined
06:56:43  * plutoniixjoined
06:56:51  * plutoniixquit (Max SendQ exceeded)
06:57:33  * plutoniixjoined
10:16:16  * joyeequit (Remote host closed the connection)
10:17:30  * joyeejoined
10:20:31  * plutoniixquit (Quit: Leaving)
10:22:08  * joyeequit (Ping timeout: 246 seconds)
10:56:47  * rmcilroyquit (Ping timeout: 255 seconds)
11:08:35  * rmcilroyjoined
11:25:10  * mylesborinsquit (Quit: farewell for now)
11:25:41  * mylesborinsjoined
11:26:06  * joyeejoined
11:30:41  * joyeequit (Ping timeout: 260 seconds)
11:41:10  * joyeejoined
12:29:54  <trungl-bot>Tree closed by [email protected]: closed (http://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20debug/builds/13871 from f75748cf4ca696fff268787068419604815f8090)
12:46:01  <trungl-bot>Tree opened by [email protected]: open
12:55:38  * sxaquit (Ping timeout: 240 seconds)
13:00:19  * sxajoined
13:19:53  * joyeequit (Read error: Connection reset by peer)
13:20:18  * joyeejoined
13:28:08  * sxaquit (Ping timeout: 240 seconds)
13:31:06  * sxajoined
13:55:50  * Garbeequit
13:58:26  * Garbeejoined
14:02:15  * bradleymeckjoined
14:04:25  * joyeequit (Read error: Connection reset by peer)
14:04:56  * joyeejoined
14:14:47  * unixpicklejoined
14:41:38  * bradleymeckquit (Ping timeout: 240 seconds)
15:29:15  <trungl-bot>Tree closed by [email protected]: closed (http://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20debug/builds/13878 from c45d065edc0ca83ca342cdb49e69a9203e532007)
15:46:40  <caitp>gsathya: another question, why do we allocate resolving functions so early for PromiseResolveThenableJob? it looks like we can defer that until the microtask is evaluated, which might help keep them in the young generation
15:47:07  * RT|Chatzillaquit (Quit: ChatZilla 0.9.86.1 [Firefox 2.0.0.22pre/2010030309])
15:47:22  <caitp>I dunno what the odds are of promoting resolving functions in between microtasks, but it seems possible
15:48:21  * unixpicklequit (Quit: My Mac has gone to sleep. ZZZzzz…)
15:53:50  * unixpicklejoined
15:59:10  <gsathya>caitp: previously, CreateResolvingFunctions was in js, so it was expensive to call out to it from PromiseResolveThenableJob
15:59:15  <gsathya>but we can change it now
15:59:35  <gsathya>there's no observable behvaior
16:00:14  <caitp>btw, this comes up at the moment due to porting some promise-use code to C++ to shrink the snapshot, which has grown a bit too big
16:00:34  <caitp>another alternative is to avoid inlining so many Promise calls, in invoke them as stubs instead
16:01:44  <gsathya>caitp: https://bugs.chromium.org/p/v8/issues/detail?id=5737#c6
16:02:05  <gsathya>RegExp builtins are 5x the size of promise builtins
16:02:40  <caitp>sure, but we should still try to shrink this down
16:03:13  <caitp>like, maybe PromiseResolveClosure can inline InternalResolvePromise, but AsyncFunctionAwait* probably shouldn't
16:03:47  <gsathya>agreed
16:04:15  <gsathya>i'm more comfortable with async builtins calling out to promise builtin stubs
16:04:20  <caitp>and now there's an entire other set of Async*Await methods lined up to enter the codebase, it would probably be better if Uncaught and Caught were the same function
16:04:34  <caitp>lots of room for improvement
16:05:03  * unixpicklequit (Quit: My Mac has gone to sleep. ZZZzzz…)
16:06:32  <trungl-bot>Tree opened by [email protected]: open
16:26:25  * unixpicklejoined
17:11:56  <caitp>we could also shrink Promise builtins code size massively by not allocating a separate fixed array for each component of a Promise
17:12:38  <caitp>could just allocate a single fixed array, a multiple of the number of elements needed, and fill the other fields with the hole when they can't be used
17:13:22  <caitp>then you'd still get the fast case for 0 or 1 callbacks, and much smaller code for InternalPerformPromiseThen
17:13:57  <caitp>not to mention a smaller heap graph to traverse
17:59:03  * bradleymeckjoined
18:13:20  * unixpicklequit (Quit: My Mac has gone to sleep. ZZZzzz…)
18:41:39  <trungl-bot>Tree closed by [email protected]: closed (http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20ASAN/builds/18928 from a5af7fe9ee388a636675f4a6872b1d34fa7d1a7a)
19:23:25  * joyeequit (Remote host closed the connection)
19:33:07  <caitp>with some combination of flags, or maybe regardless of flags, all user-supplied JS code seems to have a code with kind=BUILTIN, which doesn't seem to make any sense
19:46:04  <trungl-bot>Tree opened by [email protected]: open (Flake?)
19:47:33  <aklein>caitp: isn't that just because the code gets set to the CompileLazy builtin?
19:47:59  <caitp>I think it would have its proper entry set after being invoked
19:50:53  <caitp>it may well be that, but it just seems to make printing the code misleading
19:51:21  * unixpicklejoined
20:31:39  <gsathya>caitp: did you try changing AsyncFromSync builtins to call out to promise builtins stubs?
20:32:28  <gsathya>we can have some of the promise builtins be both TFS and TFJ stubs
20:32:35  <caitp>not yet, but we will see
20:33:35  <gsathya>it might be better to do just that, rather than reimplementing all this in c++ again
20:35:17  <caitp>another thing I could do is just change the C++ builtins to call the TFJ natives instead of reimplementing in C++
20:35:43  <caitp>lots of options, feel free to leave comments on the CL about it
20:35:53  <caitp>or on the bug
21:02:11  * sxaquit (Ping timeout: 264 seconds)
21:08:55  * sxajoined
21:20:50  * bradleymeckquit (Quit: bradleymeck)
21:54:16  * bradleymeckjoined
22:04:46  * unixpicklequit (Quit: My Mac has gone to sleep. ZZZzzz…)
22:24:59  * bradleymeckquit (Quit: bradleymeck)
22:48:09  * RT|Chatzillajoined
23:23:58  * sxaquit (Ping timeout: 258 seconds)
23:29:23  * unixpicklejoined
23:29:38  * sxajoined
23:30:41  * joyeejoined
23:40:05  * unixpicklequit (Quit: My Mac has gone to sleep. ZZZzzz…)
23:53:11  * unixpicklejoined