Conversation with #inferno at Thu Jul 21 12:04:20 2011 on powerman-asdf@irc.freenode.net (irc)
(15:36:26) Maxdamantus [~Maxdamant@203-97-238-106.cable.telstraclear.net] entered the room.
(15:40:21) Maxdamantus: Reading http://www.vitanuova.com/inferno/papers/dis.html ; it talks about byte addresses and alignment .. why exactly is that important? 
(15:41:34) Maxdamantus: I might've gone through too quickly, but I couldn't see any instructions for byte-indexing operations.
(15:43:02) Maxdamantus: (allowing byte-orientated data manipulation doesn't make sense to me for a precise garbage collected system)
(15:44:22) Maxdamantus: The closest I noticed are movm[p], which don't seem to really rely on actual byte copying; just object copying.
(15:45:04) Maxdamantus: precise-garbage-collected*
(15:54:25) xyz__ [~xyz@p578F2D1A.dip.t-dialin.net] entered the room.
(15:55:47) mjl-: Maxdamantus: which sections are you referring too?
(15:56:40) Maxdamantus: Near the top: 2 Addressing Modes
(15:56:55) Maxdamantus: "Memory Organization"
(15:58:02) xyz_ left the room (quit: Ping timeout: 258 seconds).
(15:58:59) mjl-: i see
(15:59:06) mjl-: that section reads a bit strange to me too
(15:59:15) mjl-: there is no direct byte addressing of random pieces of memory
(15:59:18) mjl-: in dis
(15:59:27) Maxdamantus: I also can't see any reason why it would think of types as arrays of bytes.
(15:59:39) mjl-: true, you wouldn't really
(15:59:46) mjl-: i think this is a thought about the implementation shining through
(16:00:09) Maxdamantus: Mm, possibly.
(16:00:13) mjl-: perhaps its useful for if you're writing a jit compiler for a new arch
(16:02:10) Maxdamantus: Are there any known attempts of implementing Dis on top of a reasonably high-level language that already has GC, etc, like Java/JavaScript/etc?
(16:02:28) Maxdamantus: (where it would be preferable not to think about pointers as byte arrays)
(16:03:06) Maxdamantus: I'd imagine it should be straighforward, as opposed to trying to implement a machine for running C programs.
(16:03:43) Maxdamantus: (where you really do need to think of pointers as byte arrays—or at least have some way of representing them as such)
(16:04:40) mjl-: i don't know of any such effort, but i agree it shouldn't be unreasonably hard to do
(16:04:56) mjl-: well
(16:05:03) mjl-: let me think
(16:06:18) ***Maxdamantus was thinking of trying to implement Dis in JS, for fun/experimentation.
(16:06:53) mjl-: types have a pointermap, indicating which field is a pointer.  knowing what is a pointer and what not may be required for safety.  but perhaps it's just used for speed.
(16:07:01) mjl-: sounds fun!
(16:07:08) mjl-: do it :)
(16:08:41) mjl-: in case you haven't found it yet, try wm/rt for inspect .dis files
(16:08:47) mjl-: i always forget the name of that tool
(16:09:39) ***Maxdamantus tries.
(16:10:57) Maxdamantus: Mm, that looks useful.
(16:13:40) Maxdamantus: So types have a size, apparently.
(16:32:34) The account has disconnected and you are no longer in this chat. You will be automatically rejoined in the chat when the account reconnects.