Da
Show enters and exits. Hide enters and exits.
| 00:07:44 | evan | BrianRice-work: you around? |
| 00:15:50 | BrianRice-work | evan: yeah, what's up? |
| 00:16:11 | evan | you seemed quite knowledgable about SELFs 'when to JIT' |
| 00:16:17 | evan | curious what slate used/uses/will use |
| 00:16:44 | BrianRice-work | not that knowledgeable. mainly, I've forgotten too many things about it that I once knew |
| 00:17:15 | evan | such is life. |
| 00:17:17 | BrianRice-work | well, Slate's still experimenting on that front. we're hugely message-passing heavy and also using multi-dispatch, so we're not sure what'll pan out. |
| 00:17:25 | evan | ok |
| 00:17:27 | evan | just curious. |
| 00:17:45 | evan | this, like write barriers, is an area that is quite hard to get a good survey of |
| 00:17:48 | BrianRice-work | I think right now we just focus on monomorphic sends |
| 00:18:07 | BrianRice-work | yeah, write barriers don't get a lot of cross-implementation comparison |
| 00:18:30 | evan | i only inline monomorphics sends now. |
| 00:18:48 | BrianRice-work | I have a copy of Iain Craig's Virtual Machines book, and he does some comparison but he's academic so his filter is set relatively high |
| 00:19:33 | evan | hm |
| 00:19:50 | BrianRice-work | also his book on the interpretation of oo languages, which mainly focuses on semantic models |
| 00:19:51 | evan | i've got the Smith/Nair Virtual Machines |
| 00:20:09 | BrianRice-work | I've heard that one's a lot more pragmatic |
| 00:20:20 | BrianRice-work | I do have the canonical GC book, of course |
| 00:20:35 | BrianRice-work | but that was state of the art circa 10+ years ago now |
| 00:20:54 | evan | Smith/Nair is good |
| 00:20:56 | evan | but it's broad |
| 00:21:02 | slava | hi guys |
| 00:21:06 | BrianRice-work | hey |
| 00:21:08 | evan | they cover all defintions of Virtual Machine in one book. |
| 00:21:18 | evan | I want a highlevel VM book. |
| 00:21:22 | BrianRice-work | oh, right. yeah they include machine-level VMs |
| 00:21:49 | evan | yeah |
| 00:21:54 | evan | there are parts that translate |
| 00:22:12 | evan | like their coverage of machine code translation |
| 00:22:15 | evan | Rosetta style |
| 00:22:23 | evan | can be applied to highlevel VMS |
| 00:22:27 | evan | VMs, not VMS. |
| 00:22:29 | evan | if you squint. |
| 00:22:39 | BrianRice-work | yeah that has occurred to me at times |
| 00:24:42 | BrianRice-work | I think usually that JIT-focus strategies occur on a profiling-driven basis, or ought to. so very pragmatic where you drop in a few different designs and see how that works economically |
| 00:24:55 | slava | evan: what's hacking |
| 00:25:11 | BrianRice-work | GC ideally should be like this, where you decide what GC semantic fits the memory economics you want your language to have. |
| 00:25:13 | evan | oh, chillin' |
| 00:25:20 | evan | working on JIT triggers. |
| 00:28:56 | slava | I'm at a coffee shop in auckland |
| 00:29:13 | slava | uploading some photos, then going to keep working on some ffi changes |
| 01:14:14 | evan | slava: what time is it there? |
| 01:14:33 | evan | nm. |
| 14:05:27 | kurtis | Oh man I so love the Ruby community. |
| 14:06:11 | jammi | and the community loves you too? |
| 14:23:01 | kurtis | Maybe like an abusive husband :P |
| 14:34:34 | jammi | so who's the nagging bitch? |
| 14:42:15 | kurtis | Hahaha |
| 15:39:14 | kurtis | OK, a bit of a newbie to this: I know that Rubinius is in C++, MacRuby in Objective-C, and MRI in C. What is that called exactly? |
| 15:49:39 | rue | What is what? |