| 00:39:04
| * perezd | joined |
| 00:39:32
| <perezd> | so I've been writing libraries for nodejs using crockford's module pattern technique. Is this considered inefficient to the v8 runtime? |
| 00:39:53
| <perezd> | also, what about functions that define functions within them.. |
| 01:47:41
| * CodeWar | joined |
| 01:53:42
| <scientes> | perezd, umm, that is really common |
| 01:54:02
| <scientes> | perezd, because every library wraps itsself in a annonymous function to avoid namespace issues |
| 01:54:03
| <perezd> | yeah, I know it is…I was hoping people would point me to some stuff to read about it.. |
| 01:54:24
| <perezd> | I understand the hidden class infrastructure when creating new instances of a function |
| 01:54:35
| <perezd> | but does that also count for definitions of identical functions on exec? |
| 01:54:45
| <scientes> | perezd, hidden _type_, javascript doesn't have classes |
| 01:54:48
| <scientes> | only objects |
| 01:54:55
| <perezd> | I am talking about the v8 stuff |
| 01:54:58
| <perezd> | I know it doesn't |
| 01:55:04
| <scientes> | yes, hidden _types_, not classes |
| 01:55:06
| <perezd> | i thought they were called hidden classes |
| 01:55:07
| <perezd> | okay |
| 01:55:09
| <perezd> | hidden types |
| 01:55:27
| <scientes> | hidden classes would something entirely differn't |
| 01:55:45
| <scientes> | mozilla/firefox's jaegermonkey also have type inference now |
| 01:55:48
| <scientes> | *has |
| 01:55:54
| * Net147 | joined |
| 01:55:57
| <scientes> | which they call it instead of "hidden types" |
| 01:56:06
| <CodeWar> | curious if like Mozilla a single top level script is only accessed by a single thread (meaning compiler is foreground one and GC runs in the same thread) |
| 01:56:26
| <scientes> | they also have incremental GC now too |
| 01:56:34
| <scientes> | like v8 |
| 01:56:59
| <CodeWar> | on ARM the difference in perf is way too vast for you guys to worry about :-) |
| 01:57:09
| <CodeWar> | including that for IonMonkey |
| 02:03:37
| * scientes | quit (Ping timeout: 246 seconds) |
| 02:17:00
| * scientes | joined |
| 02:25:07
| * CodeWar | quit (Quit: Leaving) |
| 02:30:33
| * scientes | quit (Ping timeout: 265 seconds) |
| 02:42:41
| * scientes | joined |
| 02:54:40
| * temp01 | joined |
| 02:54:51
| * temp01 | quit (Read error: Connection reset by peer) |
| 02:59:53
| * isaacs | quit (Remote host closed the connection) |
| 03:07:20
| * isaacs | joined |
| 04:39:22
| * scientes | quit (Ping timeout: 246 seconds) |
| 04:51:49
| * scientes | joined |
| 05:00:16
| * eoh | joined |
| 06:15:03
| * isaacs | quit (Remote host closed the connection) |
| 06:58:11
| * scientes | quit (Ping timeout: 260 seconds) |
| 08:12:04
| * rendar | joined |
| 08:57:40
| * C-Man | joined |
| 09:05:04
| * TheJH | joined |
| 09:12:09
| * Net147 | quit (Quit: HydraIRC -> http://www.hydrairc.com <- *I* use it, so it must be good!) |
| 09:49:38
| * TheJH | quit (Read error: Operation timed out) |
| 10:09:20
| * TheJH | joined |
| 10:31:55
| * ycou | joined |
| 10:35:19
| * [[zz]] | quit (Ping timeout: 246 seconds) |
| 10:48:55
| * [[zz]] | joined |
| 10:54:11
| * decoder | quit (Ping timeout: 264 seconds) |
| 10:54:28
| * decoder | joined |
| 12:15:33
| * eoh | quit (Read error: Connection reset by peer) |
| 12:15:37
| * eoh| | joined |
| 14:16:52
| * C-Man | quit (Quit: Connection reset by beer) |
| 14:39:01
| * C-Man | joined |
| 15:07:40
| * isaacs | joined |
| 15:15:03
| * RT|Chatzilla | quit (Quit: ChatZilla 0.9.86.1 [Firefox 2.0.0.22pre/2009081014]) |
| 15:32:22
| * scientes | joined |
| 15:39:10
| * scientes | quit (Ping timeout: 248 seconds) |
| 15:41:05
| * scientes | joined |
| 16:06:54
| * TheJH | quit (Ping timeout: 265 seconds) |
| 16:16:06
| * scientes | quit (Ping timeout: 252 seconds) |
| 16:34:10
| * TheJH | joined |
| 16:34:12
| * TheJH | quit (Changing host) |
| 16:34:12
| * TheJH | joined |
| 17:29:16
| * TheJH | quit (Ping timeout: 244 seconds) |
| 17:47:35
| * ycou | quit (Quit: Page closed) |
| 18:43:06
| * TheJH | joined |
| 19:22:32
| * seventh | joined |
| 19:32:29
| * isaacs | quit (Remote host closed the connection) |
| 19:32:49
| * isaacs | joined |
| 20:03:25
| * mraleph | joined |
| 20:41:52
| * TheJH | quit (Ping timeout: 244 seconds) |
| 20:46:37
| * CodeWar | joined |
| 20:48:07
| * scientes | joined |
| 20:49:08
| <CodeWar> | So for Mozilla JaegerMonkey on ARM we implemented an OS optimization recently of using ThreadLocal memory. In turn the OS unicasts TLB inval and flush operations for such surfaces. This helped bring IPC up by 10~20% |
| 20:49:33
| <CodeWar> | I m wondering if Chrome can take advantage of the same too. For that to be possible, the CodeCache and heaps need to be accessed only from a single thread |
| 20:50:03
| <CodeWar> | The compiler needs to be a foreground compiler too and the GC also needs happen on the same thread (some room for relaxations here) |
| 20:50:05
| <CodeWar> | thoughts? |
| 21:19:02
| * perezd | quit (Quit: perezd) |
| 21:19:10
| * CodeWar | quit (Quit: Leaving) |
| 21:58:29
| * rendar | quit (Quit: hz preparati...sarai distrutto!) |
| 22:25:49
| * RT|Chatzilla | joined |
| 22:51:18
| * CodeWar | joined |
| 23:03:50
| * scientes | quit (Ping timeout: 252 seconds) |
| 23:05:35
| * decoder | quit (Ping timeout: 264 seconds) |
| 23:06:07
| * decoder | joined |
| 23:33:08
| * mraleph | quit (Quit: Leaving.) |
| 23:38:18
| * mraleph | joined |
| 23:38:46
| * mraleph | quit (Client Quit) |
| 23:42:40
| * isaacs | quit (Remote host closed the connection) |