Experimental IRC log happs-2008-08-28

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.

11:15:07<MarcWeber>Mmm. My vxml experiment seems to work .. in theory.. There are still some glitches when adding CDATA. But the real problem is speed. I takes 3 minutes to compile the final .hs file. The code from dtd is generated within one sec..
22:30:07<Saizan_>stepcut: you've an ircbot that uses happs-state right? have you thought about if it's usable with a plugins architecture?
22:31:44<Saizan_>happs-state wants to know all the active components statically, right?
22:36:15<stepcut>let me ponder
22:37:10<stepcut>Saizan_: do you want dynamic loading of plugins?
22:37:23<Saizan_>it would be nice
22:38:24<Saizan_>also, on a checkpoint, everything gets serialized in a single file?
22:38:40<stepcut>re checkpoint, I believe so
22:38:49<Saizan_>ACTION haven't used happs much lately
22:39:30<stepcut>a tricky part with dynamic loading of plugins is that you have to restart the state engine I suspect
22:40:14<stepcut>static plugins is easy using the current Component architecture
22:40:40<stepcut>for the most part, each plugin is just a separate Component
22:40:40<Saizan_>uhm, even without dynamic loading if i remove a plugin i must be careful that there aren't events for it in the log, i guess
22:41:12<Saizan_>(modulo english)
22:41:19<stepcut>yeah
22:41:42<Saizan_>do you know if forcing a checkpoint solves that?
22:42:03<stepcut>doubt it, it would still need to be able to deserialize the state for that component which is stored in the checkpoint
22:42:21<Saizan_>right..
22:42:55<Saizan_>so once you've a component you're married to it?
22:43:02<stepcut>but, with the components being somewhat isolated from each other, it seems like there should be a way to just ignore a component
22:43:33<Saizan_>yeah
22:44:11<Saizan_>i guess i need to read the source :)
22:45:30<stepcut>dynamic plugins would be nice
22:45:43<stepcut>it would also be nice if you could run more than one state engine in the same application
22:45:58<stepcut>but HAppS-State uses some global variables :-/
22:46:05<Saizan_>:\
22:47:35<Saizan_>so that query/update can communicate with the central system?
22:48:27<stepcut>yeah
22:49:46<Saizan_>i wonder why they didn't use a monad instead
22:50:13<stepcut>Saizan_: dunno, they clearly aren't afraid of using monads and monad transformers
22:50:48<Saizan_>i hope it wasn't for ease of use
22:56:35<stepcut>I don't think you would want the plugin for a state automatically removed when it is not loaded, just ignored, right?
22:57:05<stepcut>though, it might be tricky to ignore it if you don't have the serialization code for the types lying around
22:58:30<stepcut>also, with sharding, maybe the state for different components will be in different state files ?
23:07:31<Saizan_>yeah, ignored, i might reload it later
23:08:19<Saizan_>lemmih said something along the lines that sharding will be like a distributed hash table
23:08:53<stepcut>:-/
23:09:35<Saizan_>you can ignore component it if you serialize size boundaries in the checkpoint, for events you may just keep them in the log..
23:09:59<Saizan_>"components"
23:10:36<stepcut>yeah
23:12:37<Saizan_>happs-state also makes it hard to modify the state with some external program.. maybe we want a [de]serializer for checkpoints
23:18:05<stepcut>I think you should be able to leverage the existing serialization stuff to write out all the data to a external file, perhaps as XML
23:19:03<Saizan_>yeah, but if it's not done by the central system i lose ACID
23:19:19<stepcut>right, but I think checkpoint and export should be separate
23:19:29<stepcut>though they might share some common code
23:21:11<stepcut>for a single component, ACID is not an issue, just have a query function, allState :: Query MyComponent MyState ; allState = ask, which gets you the whole state at once.
23:21:39<stepcut>not sure about multiple components
23:21:47<Saizan_>yeah, but i want to get it back inside after i've changed it externally :)
23:22:01<Saizan_>without losing events in the meantime
23:24:01<stepcut>hmm

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