Experimental IRC log happs-2007-11-03

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.

18:44:25<kc5tja>hmm...instructions for HAppS located at http://happs.org/HAppS-Begin/public/index.html appear to no longer be accurate.
18:45:52<kc5tja>I have GHC 6.6, and I've had to manually install Unix-2.0, but it still complains about not finding the symbol "lstat" when I try to invoke MyAppDir.
18:45:59<kc5tja>I used: ../sp runhaskell -i../unix-2.0 -ihaskell haskell/Main.hs
18:46:11<kc5tja>where I've installed sp in the previous directory, as I have with Unix-2.0 as well.
18:46:28<kc5tja>Anyone have any ideas, or can point me to more up to date documentation?
18:55:54<Saizan_>tried installing unix with cabal?
19:07:11<kc5tja>Yes.
19:07:15<kc5tja>That's how I built it.
19:40:23<sorear>kc5tja: http://hackage.haskell.org/trac/ghc/ticket/1086
19:40:46<sorear>so either upgrade GHC, or compile your program with -fvia-C
19:51:03<kc5tja>But, I -already- have 6.6.1.
19:51:13<kc5tja>Looks like -fvia-C is required no matter what.
20:10:15<kc5tja>Recompiling GHC just to be safe...
21:03:37<kc5tja>Good lord -- I forgot how long it took Haskell to recompile. :)
21:36:15<stepcut>kc5tja: just need a 16 core machine like dons used to have access to :)
21:36:34<kc5tja>Yeah, no kidding. :-)
21:36:51<kc5tja>I just hope I can get HAppS running easily after the recompile.
21:37:00<kc5tja>ACTION really doesn't want to fudge with it for too long.
21:37:27<kc5tja>Are there any plans on offering, um, more comprehensive documentation than what's just on happs.org? ;D
21:38:02<stepcut>I wonder if HAppS will be updated to require 6.8.1. There are some issues with incoherent/overlapping instances that might be better solved with the the new associated type synonym stuff
21:38:56<Saizan_>do you think they would help?
21:39:04<Igloo>stepcut: Does HAppS really use incoherent instances anywhere?
21:41:03<stepcut>Igloo: not sure, maybe just overlapping, let me check
21:41:21<Saizan_>btw, i hope 6.8 doesn't report the wrong class anymore when you e.g. don't have a StartState instance for a type
21:42:05<Igloo>Saizan_: I don't remember seeing that reported, so I have no idea if it's fixed or not
21:42:51<stepcut>looks like HAppS-State/HApps-Data have Overlapping and *Undecidable* instances
21:43:10<kaol>eek, and I wanted to run happs on Debian stable... time to backport ghc, when 6.8 is released?
21:43:21<Igloo>Right, that's fine
21:43:23<stepcut>I am not especially pleased with the way it undecides, toXml ([1,2] :: [Int])
21:44:08<stepcut>but, it appears there is a reason for it :)
21:45:46<stepcut>what happened to orderBy for IxSets, btw
21:47:28<Saizan_>replaced with groupBy iirc
21:53:19<stepcut>groupBy seems to use Map.toList instead of Map.toAscList. Does Map.toList guarantee any particular ordering ?
21:54:12<sjanssen>stepcut: the docs don't guarantee it, but the implementation does
21:54:53<stepcut>perhaps it would be better to use toAscList then ?
21:56:33<stepcut>Map.toAscList and Set.toAscList
22:02:37<kc5tja>Looking at this site, http://blog.moertel.com/articles/2006/10/18/a-type-based-solution-to-the-strings-problem , has there ever been a recorded instance of a successful XSS or SQL injection when using the type system to protect against un-escaped strings?
22:03:06<stepcut>If I currently have, ok (fs :: IxSet Fantasy), and I want to instead show things in reverse cronological order, this is what I have:
22:03:09<stepcut> ok $ (Elem "ixSet" $ concatMap (toXml . snd) $ reverse $ (groupBy fs :: [(FantasyId, [Fantasy])]))
22:04:05<stepcut>which seems very non-obvious -- is there a better approach ?
22:05:13<stepcut>kc5tja: I think the first question is, has anyone ever deployed a site which used the type system to protect against un-escaped strings ?
22:05:24<Saizan_>mmh, sortBy x ixset = map snd (groupBy fs `asTypeOf` [(x,undefined)]) ?
22:06:06<Saizan_>s/map/concatMap/
22:06:09<kc5tja>stepcut: I know Google (at least in the project I am in) uses type-based systems, but using Java.
22:09:17<stepcut>Saizan_: yeah, that seems to work, maybe we can submit a patch to add orderBy and rOrderBy back into IxSet.hs
22:09:55<stepcut>I think most HAppS apps are free of SQL injection attacks ;)
22:10:22<Saizan_>well orderBy was exactly like groupBy no?
22:10:51<kc5tja>stepcut: I'm asking for different reasons.
22:11:01<kc5tja>stepcut: It has nothing to do with HAppS.
22:12:38<stepcut>kc5tja: I know of no recorded successful instances of SQL or XSS inject when using the type system. But I personally know of virtually no successful attacks period
22:12:57<stepcut>I know there are lots of them, I just don't know any details
22:13:16<stepcut>Saizan_: is orderBy different than sortBy ?
22:24:38<Saizan_>stepcut: orderBy was implemented the same as the current groupBy, just with a confisingly too general type signature. so yes orderBy was different then my sortBy
22:25:05<Saizan_>*than
22:25:42<Saizan_>and maybe we should call whatever goes in IxSet orderBy since that's the SQL nale
22:25:44<Saizan_>*name
22:26:33<stepcut>Saizan_: ah, well I think your sortBy would be useful as a standard IxSet library function
22:27:07<stepcut>Saizan_: I did not use the older orderBy so I assumed it was similar to your sortBy
22:28:05<Saizan_>so you like the phantom parameter?
22:30:34<stepcut>dunno
22:30:48<stepcut>I don't like either way :)
22:31:11<stepcut>but I think the fantom param might be better
22:31:30<stepcut>is this how you intended it to be used: sortBy (undefined :: FantasyId) fantasies
22:32:10<Saizan_>yes
22:33:19<stepcut>lunch, bbiab

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