Conversation with #inferno at Thu Jan 27 21:39:01 2011 on powerman-asdf@irc.freenode.net (irc)
(22:23:49) bvalek2 left the room (quit: Quit: Page closed).
(22:25:17) Fish- [~Fish@9fans.fr] entered the room.
(00:19:13) Fish- left the room (quit: Quit: So Long, and Thanks for All the Fish).
(00:42:09) caerwyn [~chatzilla@67-133-179-109.dia.static.qwest.net] entered the room.
(01:15:39) idr0 left the room (quit: Remote host closed the connection).
(02:40:43) KillerX left the room (quit: Quit: KillerX).
(02:41:22) KillerX [~anant@nat/mozilla/x-hjhfssvjrixntxcw] entered the room.
(03:27:30) base2design [~base2desi@97-80-161-40.dhcp.gwnt.ga.charter.com] entered the room.
(04:03:34) tswett left the room (quit: Quit: Changing server).
(04:04:02) Warrigal [ihope@thay.Stanford.EDU] entered the room.
(04:04:52) Warrigal left the room (quit: Client Quit).
(04:16:49) caerwyn left the room (quit: Ping timeout: 240 seconds).
(06:37:14) base2design: so i was reading through a few research papers talking about styx/9p2000 and one made a statement saying "unfortunately the styx protocol forbids a file from being opened by more than one client at a time". is that true? i can't find any evidence to prove it...
(07:37:54) bvalek2 [c3e41404@gateway/web/freenode/ip.195.228.20.4] entered the room.
(08:08:17) robot12 [~kazzhilki@proxy10.ts.fujitsu.com] entered the room.
(09:05:33) idr0 [~idr@e179146032.adsl.alicedsl.de] entered the room.
(09:23:50) mjl-: base2design: no, that's not true
(09:23:56) mjl-: perhaps there was more context to that statement?
(09:24:38) base2design: it's the paper titled "Using the Styx Network Protocol in Event Driven Systems"
(09:25:43) base2design: by Shannon,Freeman,Bailey from University of York UK.
(09:26:13) base2design: it didn't mesh with my understanding.
(09:27:31) base2design: i've been messing around with pub/sub protocols (not on inferno) looked at this doc to see what other had messed with
(09:53:05) mjl-: just looked at the paper.  it's certainly not true now.  perhaps it was a limitation of a library they used when they wrote that.
(09:56:22) base2design: the whole multipipe thing looked strange to me.
(09:59:15) mjl-: same here
(10:00:15) base2design: it did cause me to poke around inside the tail.b in inferno... a quick test seems to indicate that at least from "real" (host os files) use a sleep delay to see if there's more file to read...
(10:00:40) base2design: mqtt is the pub/sub protocol i've looked at.
(10:01:16) base2design: yuck. i have to sleep now.
(10:01:35) mjl-: tail.b has to do that, there is no way to get notifications on changes.
(10:02:03) mjl-: there have been people writing something to use linux file system change events (inotify or what is it called?)
(10:02:11) mjl-: but that's not portable of course
(10:02:15) base2design: that's right... inotify.
(10:02:45) base2design: surely that limitation is just on hosted os based files, right?
(10:03:16) base2design: seems like you should be able to do a "blocking" read from a styx-based file though?
(10:03:28) mjl-: on the paper that uses styx:  they seem to implement styx in the fpga.  that often seems overkill to me, why not do something as simple as possible in the hardware, and write a simple styx program that translates between the hardwares (simple) interface and styx.  perhaps it's because i find software easier than hardware...
(10:03:43) mjl-: base2design: no, it's not possible in native inferno either.
(10:03:53) mjl-: a read at the end of the file just returns 0 bytes immediately
(10:04:24) mjl-: on-disk filesystems "should" have a synthetic file for file sytem changes
(10:04:26) base2design: hmmm... pub/sub may have to do the stupid polling stat thing though. yyuck
(10:04:33) mjl-: but i don't think current file systems do that.
(10:04:56) mjl-: i did have a proposal for a new filesystem i was working on to get such events.  but that filesystem isn't finished yet.
(10:05:26) mjl-: base2design: you could interpose a styx server that intercepts reads & writes
(10:05:39) mjl-: that's not very clear
(10:06:20) mjl-: i mean:  a styx server that sits between writes & readers of files.  and that blocks reads until a writer writes something new, and only then lets reads return the new data.
(10:06:35) mjl-: that does mean you have to control who is writing, i.e. "capture" them in your interposing file server
(10:06:42) mjl-: i don't know of any such server existing currently though
(10:07:45) base2design: i need to think about it more. thanks for the ideas. gotta "wake up" in a few hours so i suppose i should try to sleep now ;)
(10:07:47) base2design: night!
(10:07:55) mjl-: sleep well! o/~~
(10:43:20) idr1 [~idr@e179151169.adsl.alicedsl.de] entered the room.
(10:43:27) idr0 left the room (quit: Ping timeout: 240 seconds).
(11:46:10) Gegemon [~ynv@mx1.airis.ru] entered the room.
(13:52:07) idr1 left the room (quit: Remote host closed the connection).
(13:54:13) idr [~idr@e179151169.adsl.alicedsl.de] entered the room.
(13:57:27) idr left the room (quit: Remote host closed the connection).
(13:59:46) idr [~idr@e179151169.adsl.alicedsl.de] entered the room.
(14:06:32) Fish left the room (quit: Ping timeout: 276 seconds).
(15:17:51) acmeuser [~acmeuser@125.71.210.58] entered the room.
(15:17:58) acmeuser left the room (quit: Remote host closed the connection).
(15:30:38) Fish [~Fish@exo3753.pck.nerim.net] entered the room.
(15:42:35) bvalek2_ [c3e41484@gateway/web/freenode/ip.195.228.20.132] entered the room.
(15:43:22) bvalek2 left the room (quit: Ping timeout: 265 seconds).
(15:43:52) bvalek2_ is now known as bvalek2
(16:14:28) The account has disconnected and you are no longer in this chat. You will be automatically rejoined in the chat when the account reconnects.