Experimental IRC log happs-2007-11-21

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.

00:13:45<alexj>cedrickshock: runghc does really weird stuff with memory. the compiled version is order-of-magniture or two smaller/faster.
00:14:34<cedricshock>alexj: So I should be following the cabal instructions? (That's what I'm doing right now).
00:14:58<alexj>I never use cabal.
00:15:09<alexj>i use searchpath.
00:15:56<cedricshock>That's what I was doing before (sp runhaskell -ihaskell haskell/Main.hs), how do I compile instead of runhaskell?
00:18:49<alexj>ghc
00:19:13<alexj>s/runhaskell/ghc
00:19:21<alexj>or use HAppS-Deploy
00:19:55<alexj>argh. shapr took away that documentatiojn!
00:20:02<alexj>needs to be put back.
00:20:11<alexj>and put ahead of the cabal stuff.
00:20:29<cedricshock>Thanks. I downloaded and made http://searchpath.org/searchpath/SearchPath.hs again, but I still need the -X arguments to make ghc happy.
00:22:00<cedricshock>Or rather to make runhaskell happy, ghc is obviously being run with -fglasgow-exts
00:23:17<alexj>cedrickshock, I just pushed a new version of searchpath that also adds -fglasgow-exts for runghc
00:23:24<alexj>does not do that for runhaskell
00:23:35<alexj>runhaskell is wrong.
00:23:40<alexj>these docs need fixing.
00:23:46<cedricshock>Ok
00:24:47<shapr>What did I take away?
00:25:37<cedricshock>Excellent, sp added -fglasgow-exts to runghc.
00:31:32<cedricshock>Everything works (albeit slowly with runghc). sp runs the command
00:31:33<cedricshock>runghc -ihaskell -i.haskell_cache/http_searchpath.org-default.map.cache -fglasgow-exts haskell/Main.hs
00:32:14<cedricshock>But ../sp ghc -ihaskell haskell/Main.hs runs:
00:32:37<cedricshock>ghc -ihaskell -i.haskell_cache/http_searchpath.org-default.map.cache -fglasgow-exts haskell/Main.hs
00:32:37<cedricshock>haskell/Main.hs:3:0:
00:32:37<cedricshock> Failed to load interface for `Types':
00:32:37<cedricshock> Use -v to see a list of the files searched for.
00:43:20<alexj>you may need to wipe your *.o files.
00:43:25<alexj>you also might need to do ghc --make
00:50:09<cedricshock>Whoa! What on earth is HAppS.Server.Facebook?
00:51:01<alexj>the beginnings of a facebook lib for happs
00:51:21<alexj>for people to make Facebook apps.
00:52:16<alexj>once we get our release together we will get a link to HAppS from the facebook developer homepage.
00:52:48<cedricshock>Facebook would be a bit ugly in terms of IO, all those restfull calls back to facebook's servers...
00:53:15<cedricshock>alexj: Thanks. ghc --make pays the bills.
00:53:52<alexj>cedrickshock, it actually feels pretty natural when I am building facebook apps with it.
00:54:14<alexj>the FaceBlog.hs file in HAppS-Begin is an example.
00:56:02<cedricshock>The compiled server is much better. ~4 Mb of memory and 0.7% CPU
01:01:17<cedricshock>Is there anything I can do with HAppS-Begin out of the box other than look at "hello world"?
01:02:50<alexj>hmm someone modified happs-begin not to do anything.
01:02:54<alexj>it used to provide a basic blogging service.
01:04:06<cedricshock>It seems like HAppS could use a lot of help in the documentation department.
01:04:29<alexj>cedrickshock: if you would, that would be great.
01:04:29<Saizan>cedricshock: main = stateMain in haskell/Main.hs, rm -fR _local, recompile, rerun
01:06:31<cedricshock>Saizan: Since I'm writing down everything you tell me as instructions, what's the rm -fR _local for? To delete the current state?
01:12:22<cedricshock>Saizan: using stateMain nets me this fun error from the server: "Server error: XXX currently broken"
01:15:09<Saizan>ah, stateMain was the blog example, but it may be broken by recent changes
01:15:58<Saizan>and yes rm -fR _local is to delete the current state, since it will be incompatible with the one used by stateMain
01:20:09<Saizan>none of HaXml,hslogger,Crypto,NewBinary built without tweaking the .cabal file with 6.8.1, or are there updated packages somewhere?
01:20:42<cedricshock>I found the line at fault. "blog = error "XXX Currently broken". I wonder why this was added...
01:28:56<cedricshock>Aww, now I got real errors. After fixing the ocntext stack size in the bloq I got: Server error: Couldn't find a handler for type: GetRecent
01:41:30<Saizan>compiling HAppS.Server.Facebook is killing my poor old machine
02:28:55<cedricshock>What does "Couldn't ind a handler for type: GetRecent" mean?
02:32:16<cedricshock>s/ind/find
04:45:22<perspectivet>how are the repositories looking?
04:47:44<Saizan>have they moved?
04:49:29<perspectivet>Hey Saizan, which timezone are you in?
04:49:55<perspectivet>I thought they weren't building
04:51:06<Saizan>it's 6 am here
04:51:39<perspectivet>Italy?
04:51:45<Saizan>ah, no, now they build, with 6.8.1, you need to tweak the cabal file of some deps though
04:51:48<Saizan>yep
04:52:07<perspectivet>Have you been hacking all night?
04:52:41<Saizan>i was messing around with continuations, yes
04:53:43<perspectivet>hardcore
04:57:18<Saizan>yeah, i still find them quite unintuitive
04:58:27<perspectivet>is this directly for happs or are you exploring outside of happs?
05:01:00<Saizan>well very indirectly, i was looking for a safe way to mix the Get monad and reading from a socket, i'd like that interface for future implementations of protocols like spread
05:01:30<Saizan>in reaceive_internal i return a Get (IO Raw) to work around this
05:04:09<perspectivet>interesting
05:10:15<Saizan>btw, do you have a test program for hspread?
05:10:33<perspectivet>yeah it's about 3 lines long so far
05:11:58<perspectivet>doing most of stuff in ghci right now.
05:15:14<perspectivet>I'm working on getting the join stuff working right now.
05:25:02<perspectivet>I basically want to get functionality equivalent to the spuser program working from ghci
05:25:26<perspectivet>talking to the spread daemon
05:25:33<perspectivet>receiving messages
05:25:35<Saizan>k
05:25:48<Saizan>join still kills the connection?
05:25:48<perspectivet>currently trying to figure out why join isn't working
05:25:52<perspectivet>yeah
05:27:18<perspectivet>next step is to set some breakpoints in the spread daemon and find out how it's breaking
05:28:40<perspectivet>compare spuser to our (mostly your) stuff
05:28:51<perspectivet>do you want the privatename patch?
05:31:40<Saizan>yup
05:44:12<perspectivet>ok, so I just have to send you the numbered .gz from _darcs/patches?
05:44:34<perspectivet>haven't done the whole offline patch with darcs before
05:45:15<Saizan>after darcs record you can darcs send -o file.patch
05:47:12<Saizan>if you just use darcs send it should ask for a email address and send it for you
05:51:28<perspectivet>where file.patch is the .gz file or the darcs-record-0 file?
05:54:32<Saizan>file.patch will be created by darcs send
05:54:54<Saizan>i should have used "foo"
05:57:09<perspectivet>ok, so I did a "darcs send -o file.patch"
05:57:17<perspectivet>not sure if it emailed it back to you or not
06:00:07<Saizan>no, in that case you should have a file.patch in your current directory and you should mail that
06:00:40<perspectivet>ok, that makes sense, given the lack of delay I saw in command completion
06:03:14<perspectivet>sent
06:14:10<Saizan>ok
07:16:21<perspectivet>Saizan: you around?
07:16:26<Saizan>yes
07:17:38<perspectivet>I just traced through the spread server reception of the join message and it looks like the received message is a kILL_MESS
07:17:44<perspectivet>explains why things are dying.
07:18:19<perspectivet>haven't looked at what things look like at coming out of this side, but that'll have to wait until tomorrow.
07:18:35<perspectivet>things being the session
07:19:02<Saizan>weird
07:19:36<Saizan>i don't see how it could get out as a KILL_MESS
07:20:31<Saizan>ACTION checks the bitmasks
07:20:56<perspectivet>yeah, I couldn't see anything obvious
07:21:26<perspectivet>the first word32 comes through as 0x800400a0
07:26:48<Saizan>that's indeed equal to setEndian kILL_MESS
07:29:08<perspectivet>yeah, it's odd
07:29:17<perspectivet>that's the only part that's out of line
07:29:26<perspectivet>the rest of the message matches the spuser version
07:31:06<Saizan>there's no way disconnect could be called by your code right? (maybe used in a bracket like i was using hClose)
07:32:12<perspectivet>you mean from my test program?
07:34:18<perspectivet>I'm working from ghci
07:34:24<perspectivet>I execute:
07:34:34<perspectivet>(m,c) <- connect dc
07:34:49<perspectivet>where dc is my default Conf
07:34:52<perspectivet>then:
07:35:24<perspectivet>join (fromJust (makeGroup "blah")) c
07:36:10<perspectivet>the server receives the kILL_MESS shortly there after
07:40:00<Saizan>and if you try to send a message to yourself instead?
07:43:46<perspectivet>just going to dump the message as I send it.
07:44:51<Saizan>well, testing a Joining message with sendable it doesn't give that serviceType
07:45:42<perspectivet>ok
07:48:33<perspectivet>I'll take a look at that tomorrow if you can't find the problem
07:48:40<Saizan>ok, fine
07:54:21<Sizur>shapr: how can you compile happs for 6.8.1 when the required hslogger doesn't compile?
13:07:49<Sizur>i propose to include all the deps in the repos.
13:12:10<Sizur>i cannot build syb-with-class that HAppS-Data requires :/ all 3 versions
16:00:26<Saizan>well, we've the repo of syb-with-class
16:01:18<Saizan>http://happs.org/HAppS/syb-with-class
16:09:51<Sizur>thanks. where could i find that?
16:13:34<Saizan>apparently asking here :)
16:17:25<Sizur>do you have one for Crypto aswell?
16:24:07<Saizan>no, but for Crypto is easy to update the .cabal, i can link it to you if you wish
16:24:21<Sizur>yeah, i'm building it now
16:24:44<Saizan>ah, so you done that yourself?:)
16:24:51<Sizur>ACTION hopes cabal-install will be finished soon
16:25:24<Sizur>yeah, i wish all packages are updated instantly by authors and constantly :P
16:25:59<Sizur>of course i'm excluding myself from that rule
16:26:51<Saizan>well sendimg mails bugging the mantainers should help :)
16:29:16<Sizur>btw, since there is an ear to hear: why not include all deps in the repos?
16:29:53<Sizur>one pull should be enough to build the whole thing if all you have is ghc
16:36:30<Saizan>well, because we don't mantain things like Crypto or HaXml, we just use them
16:39:45<Sizur>but we can include a compile working version in the repo that needs it
16:40:22<Sizur>as a lib
16:48:08<Sizur>all i have from running Begin is "hello world" instead of a blog example, is that the expected behavior?
16:48:45<Saizan>yes
16:50:19<Sizur>thanks
16:53:02<Sizur>now i need to figure out how to dispatch urls :P
17:00:07<Sizur>i love to see xhtml combinators in use! :D
17:03:52<Sizur>we need a wiki
18:24:36<Saizan>perspectivet: it was an endian related bug
18:25:12<Saizan>perspectivet: now it seems to work, even if i get an endianMismatch when messages come back to me
18:28:45<perspectivet>ok, cool
18:28:54<perspectivet>so join and send are working now?
18:31:06<Saizan>yes
18:31:15<Saizan>but i don't get membership messages..
18:31:38<Saizan>i think there's an overall endian problem
18:38:08<Saizan>uhm, no, just handled some bits the wrong way..
18:38:28<perspectivet>did you find the problem?
18:41:25<Saizan>about membership messages yes
18:41:36<perspectivet>ah, cool
18:42:00<perspectivet>so you're sending join and receiving the membership reply?
18:42:05<Saizan>yes
18:42:15<Saizan>pushed fix
18:42:18<perspectivet>Sweet
19:12:54<perspectivet>Saizan: hmmm. Things are working pretty good
19:13:17<perspectivet>I'm getting messages reporting other users joining groups I've joined
19:13:35<perspectivet>Just about to test my own non-join send
19:14:09<perspectivet>disconnect and reconnect is working inside a ghci session
19:24:32<perspectivet>Saizan: I can send and receive custom message without problems.
19:25:09<perspectivet>only tested with two users in a group joined by both
19:25:19<perspectivet>but things are looking pretty good
19:30:28<Saizan>i'm a bit puzzled by the endianMismatch = True when you get your own mess back
19:30:41<Saizan>does it happen to you too?
19:31:54<perspectivet>when I get my group membership message back you mean?
19:32:22<Saizan>no, if you send a message with selfDiscard = False
19:33:21<perspectivet>that's the outdiscard flag?
19:33:33<perspectivet>in OutMsg?
19:33:41<Saizan>let me check
19:33:59<Saizan>yes, that
19:34:09<perspectivet>what exactly is that supposed to do?
19:35:13<Saizan>well if it's True you won't get a copy of the message back among the other regular messages for the groups you are sending this
19:35:37<Saizan>if it's False you get one like all the other hosts
19:35:57<perspectivet>ah, right, makes total sense to have something like that if I think about it.
19:36:38<perspectivet>I didn't seem to get a message back using outDiscard = False...
19:37:19<perspectivet>but I did get an inEndianMismatch=True in the message I sent to a my group from user.c
19:38:57<perspectivet>ah, actually I did get a message back with outDiscard=False I just didn't display it, my bad
19:39:05<Saizan>ah, ok
19:40:45<perspectivet>ok, so yeah inEndianMismatch=True on both the message i receive from user.c and on the outDiscard=False returned message
19:41:56<Saizan>yup, while spuser reports endian 0
19:42:59<perspectivet>yes I have that as well
19:46:09<Saizan>i indeed had to use reverse to make spread get the right serviceType
20:35:45<sizur>does this channel have a log?
20:36:58<shapr>yes
20:37:19<shapr>http://tuukka.iki.fi/tmp/logindex
20:37:29<sizur>i need to check my conversatin from earlier today, thanks
21:50:02<alec>are there haddock docs online? I'm installing from darcs and my haddock (also from darcs) doesn't complete on -Server, -IxSet, and -State
21:55:41<shapr>HAppS has much magic.
21:55:44<shapr>alec: I don't think so.
21:55:50<shapr>alec: I'm not sure the haddocks build right now.
21:55:55<shapr>But I can put that on the issue list.
21:57:59<alec>shapr: thanks
21:58:27<alexj>a bunch of changes that lemmih is working on
21:58:35<alexj>will make the magic much less scary.
22:09:40<sizur>what does this mean? http://hpaste.org/4029?lines=true
22:21:07<sizur>i'm tyring to bundle up the whole happs together with deps in one source for my project. and i'm stuck at the above stated compilation error
22:21:35<sizur>the first one is with the old HaXml, and second is with the new version of it
22:25:32<alexj>you are using cabal install.
22:35:13<sizur>alexj: nope, i'm doing a ghc --make
22:35:24<sizur>on a barebones Main
22:40:46<alexj>then yes right now happs is only compatible with the old haxml.
22:41:04<alexj>i have on my agenda to eliminate the dependency entirely, but we haven't gotten there yet.
22:41:19<alexj>want to use oleg's xml stuff.
23:11:06<sizur>alexj: but the old dep is producing that kind error for me
23:12:23<alexj>happs-begin was buildng with haxml earlier today.
23:13:04<sizur>i just cut a few lines from the Main of the Begin into a separate Main, not good?
23:15:42<sizur>like this: http://hpaste.org/4032?lines=true
23:33:09<sizur>can we move all extentions and options to files instead of .cabal?

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