02:03:13  * bnoordhuisquit (Ping timeout: 240 seconds)
03:05:46  * faceleg|workquit (Ping timeout: 258 seconds)
03:09:57  * bnoordhuisjoined
03:14:35  * bnoordhuisquit (Ping timeout: 258 seconds)
03:56:32  * jazz264joined
04:30:09  * C-Manquit (Quit: Connection reset by beer)
07:53:44  * Frippejoined
09:22:17  * Net147joined
09:22:34  * Net147quit (Client Quit)
09:59:14  * jazz264quit (Ping timeout: 246 seconds)
10:12:11  * Net147joined
10:49:14  * stalledquit (Ping timeout: 255 seconds)
10:49:19  * jaseemabidjoined
10:58:09  * stalledjoined
11:03:59  * facelegquit (Ping timeout: 260 seconds)
11:04:09  * facelegjoined
11:23:47  * rickytaylor26quit (Ping timeout: 255 seconds)
11:23:47  * saurikquit (Ping timeout: 260 seconds)
11:23:50  * ricky26joined
11:24:33  * saurikjoined
12:11:30  * C-Manjoined
12:22:08  * stamphojoined
12:42:10  * bnoordhuisjoined
13:39:20  * c4milojoined
14:04:18  * cha0squit (Ping timeout: 258 seconds)
14:33:41  * jaseemabidquit (Ping timeout: 248 seconds)
14:41:00  * Net147quit (Quit: HydraIRC -> http://www.hydrairc.com <- Chicks dig it)
15:12:49  * Frippequit (Quit: Computer has gone to sleep.)
15:20:09  * jaseemabidjoined
15:26:47  * {V}joined
15:30:52  <{V}>http://code.google.com/p/v8/issues/detail?id=2599 suggests https://code.google.com/p/v8/issues/detail?id=2424 should be solved first, but asm.js requires a lot of "coersion" ; would v8 even make those "mistakes" with that amount (see http://jsbin.com/atokoq/2/edit ) ?
15:31:24  <guor>what do you mean by "mistakes"?
15:32:53  <{V}>2424 complains that v8 does the addition as doubles, rather than int32
15:34:26  <guor>the coersions in asm.js you are talking about are essentially signals on the type inferrer. the type inferrer pattern matches those signals to know which types to forcibly use
15:34:43  <guor>V8 doesn't match those patterns yet since those just have not been common in javascript code, so far
15:35:00  <guor>with asm.js this may change, so V8 will be optimized to act on those patterns
15:35:12  <guor>part of it is issue 2424
15:35:35  <guor>of course asm.js enables some other optimizations as well
15:41:32  <mraleph>{V}: yes it will make it.
15:41:50  <mraleph>{V}: amount does not matter.
15:42:44  <mraleph>guor except for some tricks with bounds check elimination I don't see anything that asm.js enables that normal JIT compilation framework can't achieve.
15:43:15  <{V}>I kind of assumed v8 already did most such inferences. Especially since they (asm.js users) report some of the tests they ran do better in v8 with the asm.js version of Emscripten output then the non-asm.js output
15:44:05  <guor>i guess some pieces are still missing :)
15:45:03  <mraleph>{V} it does some but there is a problem: it still allows type feedback from operations superceed what you infer statically. in this case if you add some big int32 numbers you get double type feedback for a +
15:51:03  <{V}>except the asm.js spec currently requires that you explicitely limit it to 32bit with a |0 operation if a mere int32 is the desired result.
15:54:21  * RT|Chatzillaquit (Quit: ChatZilla 0.9.86.1 [Firefox 2.0.0.22pre/2009081014])
15:54:45  <{V}>oh right. from 2424 " u = (x4 + x0) | 0;
15:54:45  <{V}>The addition remains double. "
16:06:18  <mraleph>{V} asm.js spec has nothing to do with this. JavaScript already requires x|0 to be 32-bit integer.
16:09:02  <{V}>like you say, the asm.js spec doesn't itself require x|0 to be 32-bit integer. If it is supposed to be a 32bit integer asm.js requires you to put that |0 there.
16:17:13  <luite>annoying asm.js, everyone keeps asking about it but i can't use it :p (also it feels like a rather horrible hack...)
16:19:05  <{V}>perhaps in issue 2424 the floating point addition would've been avoided if at the top of the example u = u|0; is added.
16:19:46  <{V}>luite (yes, it does, doesn't it. )
16:36:30  * carifquit (Quit: Ex-Chat)
16:37:25  * carifjoined
16:39:27  <mraleph>{V} no u = u | 0 does not affect u = (x4 + x0) | 0. double type floats from + not from variable.
16:40:46  <{V}>kay
16:40:52  <{V}>thanks, mraleph
16:41:35  <mraleph>{V} disclaimer: I am the guy who filed 2424 :-)
16:44:00  <{V}>mraleph, I might as well ask. Am I guessing correctly that the "0 2 d1238 Change i359 i to d range[-2147483648,2147483647,m0=0] <|@" type output is only available in the standalone v8 ? (i.e. not from within chromium)
16:44:19  <mraleph>{V} this one is available from chromium as well.
16:44:41  <mraleph>{V} —no-sandbox —js-flags="—trace-hydrogen"
16:45:00  <mraleph>{V} long dash is actually double dash (stupid Adium)
16:45:11  <mraleph>{V}
16:45:39  <mraleph>{V} the only thing that is not available from standalone Chromium is disassembly output.
16:46:09  <mraleph>{V} but you can get it if you build your own Chromium.
16:46:44  <mraleph>takes a while but is doable, I did it on my Mac Mini once, so you don't need a lot of power, just a lot of time :-)
16:47:29  <{V}>thanks mraleph
16:54:48  * bnoordhuisquit (Ping timeout: 264 seconds)
16:57:20  * bnoordhuisjoined
17:29:00  * bnoordhuisquit (Ping timeout: 264 seconds)
17:39:55  * facelegquit (Quit: WeeChat 0.3.9.2)
18:00:38  * c4miloquit (Remote host closed the connection)
18:01:05  * c4milojoined
18:05:25  * c4miloquit (Ping timeout: 248 seconds)
18:18:48  * Alex_Gaynorquit (Ping timeout: 252 seconds)
18:20:06  * bnoordhuisjoined
18:41:35  * Alex_Gaynorjoined
19:08:33  * c4milojoined
19:29:47  * cha0sjoined
19:29:47  * cha0squit (Changing host)
19:29:47  * cha0sjoined
20:14:26  * carifquit (Quit: Ex-Chat)
21:04:04  * [[zz]]quit (Read error: Connection reset by peer)
21:05:03  * [[zz]]joined
22:50:05  * c4miloquit (Remote host closed the connection)
22:50:30  * c4milojoined
22:55:01  * c4miloquit (Ping timeout: 248 seconds)
23:13:55  * mralephquit (Read error: Connection reset by peer)
23:36:36  * RT|Chatzillajoined