00:19:46  * mmaleckichanged nick to mmalecki[zzz]
04:01:43  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
04:06:28  * alejandr1mgjoined
04:10:22  * alejandr1mgquit (Client Quit)
04:10:36  * alejandromgjoined
05:48:39  * alejandromgquit (Quit: leaving)
12:09:22  * mmalecki[zzz]changed nick to mmalecki
13:35:08  * mmaleckichanged nick to mmalecki[away
14:46:52  * piscisaureus_joined
15:05:09  * mmalecki[awaychanged nick to mmalecki
17:06:50  * piscisaureus_quit (Read error: Connection reset by peer)
17:07:13  * piscisaureus_joined
17:54:41  <indutny_sleeping>refactored GC a little bit
17:54:45  * indutny_sleepingchanged nick to indutny
17:55:22  <indutny>169ms on stress test
17:55:24  <indutny>was 216
17:55:34  <indutny>now time to change tagging scheme for numbers
18:28:18  <creationix>:)
18:51:55  <indutny>done
18:52:08  <indutny>now add and sub are much faster
18:55:52  <indutny>ok, now candor.io
18:57:29  * piscisaureus_quit (Quit: ~ Trillian Astra - www.trillian.im ~)
19:16:31  <creationix>alright, I should work towards http_hello world
19:16:35  <creationix>and see how fast this is
19:16:45  <creationix>compare it with node, luvit, and luvmonkey
19:16:49  <creationix>(all using libuv
19:23:25  <indutny>haha
19:23:28  <indutny>would be cool!
19:24:13  <creationix>so do you expect this to be faster and v8 and/or luajit yet?
19:24:18  <creationix>s/and/than/
19:25:40  <indutny>no
19:25:44  <indutny>it's slow so far
19:26:05  <indutny>setting getting property is done in runtime, etc
19:26:32  <indutny>the only thing that should be faster than in v8 - crossing C++-candor boundary
19:26:36  <indutny>it's very thin
19:26:51  <indutny>well, literally thin
19:28:31  <creationix>ok
19:28:38  <creationix>then it may have a change in http hello world
19:28:42  <creationix>very little is done in script
19:28:49  <creationix>*chance
19:29:01  <indutny>I hope so
19:29:12  <creationix>of course luajit is really good at that boundary too
19:29:13  <indutny>anyway, I'll add runtime optimizations and inlined object access soon
19:29:16  <indutny>yes
19:29:18  <creationix>v8 just sucks at it
19:31:26  <indutny>aah
19:31:29  <indutny>I see what's happened
19:32:30  <mmalecki>I was always wondering, can't crossing this boundary be simply a `jmp`?
19:32:54  <indutny>mmalecki: no
19:32:56  <indutny>:D
19:33:13  <indutny>GC makes thing a little bit harder than it seems to be
19:33:31  <indutny>you should store marks on stack to make GC know where C++ stack start and ends
19:33:43  <indutny>and where're your language's frames
19:35:45  <indutny>mmalecki: ^
19:36:10  <mmalecki>I see. thanks for explanation :)
19:42:31  <creationix>since we don't have prototypes or any way to create fast object instances
19:42:41  <creationix>we do need a fast way to create things
19:42:56  <indutny>fast cloning
19:43:04  <creationix>works for me
19:43:10  <creationix>clone the prototype
19:43:15  <creationix>it's an instance
19:43:29  <creationix>still don't have a "this" replacement
19:43:33  <creationix>the : syntax is one way
19:45:02  <creationix>Proto = {name:"bob",greet: (self) { print(self.name + " says hi") }}
19:45:10  <creationix>instance = new Proto
19:45:14  <creationix>instance:greet()
19:45:38  <creationix>(where `new ` is a fast clone and : is syntax sugar like in lua)
19:47:19  <creationix>instance:greet() -> instance.greet(instance)
19:47:26  <indutny>sounds good
19:48:13  <creationix>and you're free to do copy on write clones or whatever you want in the vm, it doesn't have to reflect changes to the prototype after the clone has happened
19:52:00  <indutny>ok, fixed one issue
19:52:05  <indutny>hope candor.io will work
19:52:10  <indutny>heh
19:52:20  <indutny>nope
19:53:21  <indutny>another GC problem
19:53:35  <indutny>ok, I'll test it more carefully
19:53:37  <indutny>but tomorrow
19:53:39  <indutny>thank you guys
19:53:40  <indutny>ttyl
19:53:43  * indutnychanged nick to indutny_sleeping
23:27:32  * piscisaureus_joined