00:51:07 <stepkut> for acid-state Remote.. is there any form of IPC besides Sockets that would actually be worth supporting?
00:52:07 <stepkut> the methods I know of are, pipes, FIFO (named pipes), shared memory, mapped memory, message queues, and sockets
00:53:28 <stepkut> sockets really seem like the most sensible choice
00:53:38 <stepkut> especially on Unix where you can have UnixSocket files
00:54:08 <stepkut> that pretty much usurps any reason to use pipe/FIFOs .. especially since we need bidirectionaly communication
00:56:47 <stepkut> I've abstracted out the process function so that it can be used with anything.. but I think acidServer should only worry about Socket stuff
00:57:07 <stepkut> it does need hooks for allowing you to configure the Socket and add authentication in though
00:57:45 <stepkut> and we should make it easy to use ssl
20:34:01 <donri> mm_freak: anything relevant to your issues? http://www.haskellforall.com/2013/05/pipes-330-folds-and-uniting-listt-with.html
20:34:41 <donri> "This observation led me to discover the correct solution: make StateP and WriterP share their effects globally across the pipeline, instead of locally."
20:35:05 <donri> i *think* i remember you talking about the locality of one of those transformers
20:37:05 <donri> [as a problem]
20:48:32 <mm_freak> donri: i'm working on a stream processing abstraction of my own
20:49:10 <mm_freak> fewer type arguments, proof that it's a category (in agda) and sane leftover and resource management
20:49:25 <donri> sounds cool!
20:51:29 <donri> stream processing is the new blog engine is the new irc bot
21:02:56 <donri> bedtime!
21:39:03 <levi> My stream processing abstraction isn't a category, but it makes julienne fries.