Experimental IRC log happs-2007-08-13

Available formats: content-negotiated html turtle (see SIOC for the vocabulary)

Back to channel and daily index: content-negotiated html turtle

These logs are provided as an experiment in indexing discussions using IRCHub.py, Irc2RDF.hs, and SIOC.

06:54:55<vegai>shapr: how's things? :)
13:34:01<shapr>vegai: Getting closer
13:38:00<jmelesky>shapr: anything to help with aside from testing it when it's ready?
13:38:43<shapr>Well, I'm trying to write examples at the moment.
13:39:08<shapr>The repository refactoring is pretty much done, though we did find a darcs bug in the process of doing that.
13:39:22<jmelesky>nice. not the same one everyone hits, i gather?
13:39:42<shapr>Nah, we've split HAppS into bunches of repos
13:42:55<shapr>HAppS-Agents, HAppS-Begin, HAppS-Data, HAppS-IxSet, HAppS-Server, HAppS-State, and HAppS-Util
13:43:02<shapr>Oh, and SearchPath will be separate.
13:44:59<shapr>Oh, and HAppS-Deploy is a new repo that will include code to push a HAppS app to a bunch of different servers, or EC2 or whatever.
13:45:06<jmelesky>what was the bug, then?
13:45:44<shapr>If you do a change and then a rename that changes the case of a filename, it causes problems because the file ordering is then different.
13:45:45<jmelesky>i've been wondering about the EC2 stuff. is it actually handling state across multiple machines? or should i just wait for the code reveal to answer those questions?
13:45:55<shapr>It is really handling state across multiple machines.
13:46:00<jmelesky>neat
13:46:21<shapr>Yeah, I haven't tried it yet, but lemmih did it, and a bunch of cool new stuff to go along with it.
13:46:23<jmelesky>you know, i ran into a similar darcs issue, but i figured it was related to my case-sensitive FS (i'm on OSX)
13:46:41<vegai>that might actually be a killer feature.
13:46:52<jmelesky>vegai: agreed
13:47:06<shapr>The next step is even cooler..
13:47:09<shapr>sharding!
13:47:14<vegai>ACTION is afraid now.
13:47:35<shapr>Sharding is like state across multiple machines, but with segmented chunks.
13:47:51<vegai>afraid that I have to dump all my current (mal)designs and start working on HAppS again :P
13:47:59<shapr>Right now state on multiple machines means that each HAppS instance still has to hold all the state at once.
13:48:07<shapr>sharding will change that.
13:48:32<jmelesky>so you're skipping the "dedicated state store" step
13:48:52<shapr>Oh, things have changed for the better on the IO and HTTP side as well, now there's an explicit division between state updates and state viewing.
13:49:04<shapr>It's based on the UPDATE/SELECT idea that SQL uses.
13:49:24<shapr>So there's a new SimpleHTTP that replaces SimpleHTTP2, and it is *way* better.
13:49:38<shapr>It's far more elegant, much easier to use, etc
13:49:53<Saizan>gah, code breakage?
13:49:59<shapr>Much
13:50:10<shapr>But I think you'll like the improvements.
13:50:17<shapr>I'm not easily sold on code breakage, but this is very much worth it.
13:52:02<Saizan>yeah, better break code than not use good solutions
13:52:47<shapr>Alex came up with a better way to do the happs mainloop stuff, and lemmih improved that even more, it's like the first musasabi code with all the recent improvements rolled in.
13:53:30<Saizan>so we're back to a single app function?
13:53:52<shapr>Um, not exactly... it's hard to describe concisely
13:54:05<shapr>I'll just try to get everything fixed up so we can announce the new version.
13:54:26<shapr>It's really cool, and I think you'll love some parts and hate others, but that's to be expected :-)
13:54:53<shapr>The faster we get it out there, the sooner you get cool code, and we get feedback.
13:55:13<shapr>(and I want to use it for Fermat's Last Margin!)
13:55:39<Saizan>ACTION waits the release
13:56:05<jmelesky>of course, now we have to ask: how soon is soon? :)
13:56:23<shapr>I still don't know :-(
13:56:31<shapr>Whenever alex thinks it's ready to release.
13:56:39<shapr>On the good side, we nearly announced last week, so I think it's close.
13:57:10<jmelesky>sweet. i'm offline for a week starting wednesday, so it will be a nice surprise to come back to
16:45:05<reltuk>wow...this sounds like hotness...
19:23:05<vegai>yes, the hype is on :P Perhaps we should write smashing blog articles about the next release
19:23:18<shapr>hah
19:23:23<shapr>Well, I've tagged all the repos 0.9
19:23:43<shapr>But it takes seven minutes to get them right now, so I'm searching for some way to speed up the repo
19:23:58<shapr>Otherwise I'll have to blow away all the history and re-import.
19:29:24<sorear>shapr: Have you tried darcs optimize --checkpoint ?
19:29:55<shapr>just trying that
20:42:41<shapr>sorear: It worked! darcs get went from seven minutes to seven SECONDS!
20:46:59<jmelesky>nice
20:56:06<reltuk>that's a local full-repository get?
20:58:19<shapr>No, that's a network ssh get
22:18:12<shapr>ACTION loves syb-with-class xml
22:18:45<kaol>ACTION silently nags about licenses for Debian
22:20:58<shapr>er, right :-)
22:22:10<shapr>The new repos do have COPYING
22:22:34<shapr>And the new repos may be out tomorrow
22:25:02<kaol>great
22:48:34<reltuk>how am I supposed to get any work done tomorrow if I'm inclined to play with the new api all day :-p
22:51:17<shapr>wow, this is AWESOME
22:51:23<shapr>I love XML done with syb-with-class!
22:51:27<shapr>heh
23:12:22<reltuk>shapr: it just creates the serializations for you?
23:34:06<Igloo>It gives you toXml and fromXml for all types
23:38:25<Saizan>defined as?
23:42:49<shapr>$( deriveAll [d| data Tree a = Nil | Branch (Tree a) a (Tree a) deriving (Eq, Show) |] )
23:42:52<shapr>instance Default a => Default (Tree a)
23:43:37<shapr>> toXml $ Branch Nil (1 :: Int) Nil
23:43:37<shapr>[Elem "branch" [ Elem "nil" [] CData "1" Elem "nil" []]]
23:43:57<shapr>Saizan: Is that cool or what?
23:44:39<Saizan>yeah, quite cool :)
23:45:35<shapr>That's from an example I wrote earlier today.
23:46:43<Saizan>how does it decide to use CData?
23:47:05<shapr>Ah, Igloo just improved that.
23:47:31<shapr>You can specify Eq and Show at the top of the deriveAll block, and it derives those for every datatype.
23:51:04<Igloo>You can override individual instances to use CData; the default is tags

Back to channel and daily index: content-negotiated html turtle