--- Log opened Wed Jan 28 00:00:10 2009
00:41 < h_buildbot> Build for ghc-6.8.3 just succeeded. See http://buildbot.happstack.com/ghc-6.8.3/
00:46 < h_buildbot> Build for ghc-6.10.1 just succeeded. See http://buildbot.happstack.com/ghc-6.10.1/
00:47 < wchogg> It's really nice having this buildbot here.
00:50 < mae> yep :)
00:53 < mae> what about the 'defined not used' errors
00:53 < mae> someone said something about being able to see what those names are
00:53 < mae> on another chat
00:56 < wchogg> Are you talking about the ones that look mangled?
00:57 < mae> yeah
00:57 < mae> the internal error or whatever
00:57 < mae> 11:03 < Saizan> you can use -ddump-splices to see the generated code
00:57 < mae> 11:04 < Saizan> ah, you're talking about "mkUsageInfo: internal name? TextConstruct{tc aT35} ..."
00:57 < wchogg> Ah, I hadn't tried that yet.
00:58 < mae> re: 11:01 < wchogg> If anyone has any idea what's causing all the warnings in Data/Atom it'd be appreciated.  I'm not really familiar with TH.
00:58 < mae> it's all in ehre
00:58 < mae> here
00:58 < mae> http://happstack.com/irc-logs/happs-2009-01-26.txt
01:04 < wchogg> mae : So what's your target date for 0.1?
01:08 < mae> i'm hoping next week
01:08 < mae> thinking wednesdayish
01:09 < mae> earlier or later, depending on the wind
01:10 < wchogg> Cool.
01:10 < wchogg> Seems doable.  The boring cleanup work I've been doing is _mostly_ done, I think.
01:16 < mae> yep
01:16 < mae> if you can crack those weird TH errors
01:16 < mae> that would be awsome
01:17 < wchogg> I'll dig into those in the morning.
01:17 < mae> k
01:17 < mae> just pushed up some code
01:17 < mae> for ixset
01:17 < mae> i turned on all warnings
01:17 < mae> (incomplete matches was turned off)
01:17 < mae> and fixed all those warnings
01:19 < wchogg> Cool
01:21 < mae> thanks for helping out with the cleanup, couldn't have done it without you :)
01:22 < wchogg> I'm just glad to have something to keep me from worrying about hearing back from PSU.
01:25 < mae> : )
01:26 < h_buildbot> Build for ghc-6.8.3 just succeeded. See http://buildbot.happstack.com/ghc-6.8.3/
01:29 < mae> hmm
01:29 < mae> 6.8.10 :\
01:29 < mae> 5 minutes apart i guess
01:30 < wchogg> You know that's all because of Facebook & the MD5 module
01:30 < mae> hehe
01:30 < mae> i think it might go faster if the heap is larger for building lemme see here
01:30 < h_buildbot> Build for ghc-6.10.1 just succeeded. See http://buildbot.happstack.com/ghc-6.10.1/
01:37 < mae> ok bedtime for me
01:37 < mae> night!
01:41 < mae> what do you guys think about this? http://hackage.haskell.org/cgi-bin/hackage-scripts/package/SMTPClient-1.0
01:42 < mae> its not unreasonable to point to an smtp server w/ your app
01:43 < mae> should be quick an dirty enough to get rolling with website notifications
01:43 < mae> ok really going to bed now
02:21 < h_buildbot> Build for ghc-6.8.3 just succeeded. See http://buildbot.happstack.com/ghc-6.8.3/
02:25 < h_buildbot> Build for ghc-6.10.1 just succeeded. See http://buildbot.happstack.com/ghc-6.10.1/
06:50 < koeien> good to see the buildbot worked last night :)
06:58 < koeien> mae: the SMTPClient-package seems fine, we may want to put a small wrapper around it to automagically lookup the hostname of the SMTP server. also a `sendmail'-using package would be useful
12:46 < h_buildbot> Build for ghc-6.8.3 just succeeded. See http://buildbot.happstack.com/ghc-6.8.3/
12:50 < h_buildbot> Build for ghc-6.10.1 just succeeded. See http://buildbot.happstack.com/ghc-6.10.1/
13:25 < Saizan> pretty quiet today :)
13:28 < wchogg> Yeah, the buildbot is the chattiest of us.
13:45 < koeien> which is a good sign :)
13:46 < mightybyte> I just posted an interview with mae about Happstack to my blog (http://softwaresimply.blogspot.com)
13:46 < koeien> all you people are hacking on Happstack :)
13:47 < mightybyte> It's also at http://www.reddit.com/goto?id=7t4lx if anyone wants to vote it up.
13:49 < hydo> mightybyte: that redirects to the article.
13:49 < hydo> Hard to mod up :)
13:49 < mightybyte> hydo: Oops
13:50 < mightybyte> http://www.reddit.com/r/programming/comments/7t4lx/happstack_an_interview_with_matthew_elder/
13:50 < hydo> upped!
13:50 < hydo> and now lunch..
13:53 < koeien> hey guys, i know how to set-up Happstack with SSL!
14:40 < Saizan> with a proxy?
15:17 < koeien> Saizan: yes
15:20 < koeien> i'll write an article about it
15:57 < stepcut> happstack is moving on up. People are starting to bash us!
15:58 < mightybyte> How so?
15:58 < stepcut> http://www.reddit.com/r/programming/comments/7t49j/haskell_web_frameworks_reinvent_too_much/
15:58 < mightybyte> Oh yeah, I saw that.
15:59 < mightybyte> Rather ignorant if you ask me.
15:59 < stepcut> apparently we are dump for replacing apache and fastcgi with a few hundred lines of code
15:59 < stepcut> s/dump/dumb/
15:59 < stepcut> obviously trying to get three different programs from three different vendors to work together in order to acheive the same result is far supperior ;)
16:00 < mightybyte> Of course.  Many more man-years behind it.
16:01 < mightybyte> Although it mentioned HAppS, not Happstack.
16:01 < Saizan> however i'm not sure that using lazy I/O in a webserver is really robust
16:01 < stepcut> well, the part of happstack that replaces apache+fastcgi is pretty straightforword, just HAppS.Server.HTTP.Listen, etc
16:02 < stepcut> Saizan: :)
16:02 < stepcut> Saizan: the iterator stuff looks interesting
16:03 < Saizan> maybe we could use hyena as a backend :)
16:03 < stepcut> left-fold enumerators rather
16:03 < stepcut> http://www.galois.com/blog/2008/09/12/left-fold-enumerators-a-safe-expressive-and-efficient-io-interface-for-haskell/
16:04 < mightybyte> Accusing HAppS of reinventing the wheel is like accusing Google of copying Altavista.
16:04 < stepcut> :)
16:04 < mightybyte> ...true on a superficial level, but totally missing the point.
16:04 < Saizan> http://github.com/tibbe/hyena/tree/master <- tiny HTTP server using the left-fold enumerators
16:05 < stepcut> Saizan: yeah, I am adding a bug
16:05 < Saizan> to?
16:07 < stepcut> I am adding a bug to google code stating we should evaluate if it would make sense to provide an alternative to HAppS-Server that is based on left-fold enumaration
16:10 < stepcut> done.
16:11 < stepcut> how do I make something a 'Task' instead of a 'Defect'. Can only certain users do that?
16:12 < stepcut> oh, I was supposed to do it when I created the bug
16:13 < stepcut> ok, fiked that.
16:13 < stepcut> fixed that.
16:13 < stepcut> I really know nothing about left-fold enumaration except that some smart people recommend it :) No idea if happstack can use it or not.
16:18  * stepcut runs apt-get remove --purge libghc6*
16:21 < Saizan> you need to change the .SimpleHTTP interface using enumerators instead of lazy ByteStrings
16:23 < stepcut> Saizan: do you see this interfering with the use of HAppS-State in any way? They seem orthogonal to me so far...
16:23 < stepcut> it sounds like it would not be compatibly with existing SimpleHTTP code, but I don't care about that.
16:24 < stepcut> that's what makes it an alternative ;)
16:28 < Saizan> yeah, the webserver part looks completely indipendent to me
16:28 < stepcut> sweet
16:29 < Saizan> there's a ticket for removing the -state dep from -server, in fact
16:29 < stepcut> true
16:30 < mightybyte> Yeah, I spent a few minutes looking over how to do that.  It seemed a little more involved than I had hoped.
16:30 < stepcut> though, doesn't that ticket involve axeing webQuery/webUpdate? Or maybe just moving them to the intergration package?
16:31 < Saizan> the latter
16:31 < stepcut> I'm ok with the latter
16:32 < Saizan> they are just liftIO . query / liftIO . update, anyway
16:32 < stepcut> yeah
16:46 < koeien> what is the problem keeping 'm? it would provide some backwards compatibility
16:52 < Saizan> the dependency
16:54 < koeien> so people will have to cahnge webQuery into liftIO . query by themselves?
16:54 < stepcut> koeien: ideally HAppS-Server should not depend on HAppS-State, since some people might want to use mysql instead of HAppS-State
16:54 < koeien> or postgres ;)
16:54 < stepcut> koeien: no, they will be moved into a intergration/glue module that is allowed to import both HAppS-State and HAppS-Server
16:55 < koeien> ok, i didn't know of the existence of HAppS-Glue until now . forgive me my ignorance
16:55 < koeien> or whatever it's called
16:55 < koeien> :)
16:55 < stepcut> koeien: yeah, it only exists in mae's mind so far
16:56 < koeien> so the glue will provide some convenience functions to do lifting into the correct monads?
16:57 < stepcut> yeah
16:59 < stepcut> HAppS-State really requires HAppS-Data, so that dependency is ok. But HAppS-Server barely requires HAppS-State, so it is kind of annoying that you have to install the whole happstack if you just want HAppS-Server. But at the same time, it would be annoying to not have webQuery and webUpdate provide somewhere
16:59 < stepcut> so things of that nature would go in HAppS-Glue
17:00 < koeien> HAppS-State is also not dependent on HAppS-Server
17:00 < Saizan> btw, i'd move Serialize to HAppS-State and remove the dep on -Data
17:00 < stepcut> the idea is to try to keep independent modules independent, so that you can use them individually, but still provide a place for the useful things that glue stuff together.
17:01 < koeien> stepcut: that sounds like a good idea
17:01 < stepcut> Saizan: is that all that HAppS-State uses from HAppS-Data ?
17:01 < koeien> Saizan: what's left of HAppS-Data then?
17:01 < Saizan> koeien: all the Xml and generics stuff
17:01 < stepcut> koeien: gSet, gFind, etc. It would still be used by IxSet I think?
17:02 < koeien> i'm not too familiar with the codebase as you may have noticed by now. plan to become more familiar in the future though
17:03 < Saizan> stepcut: yes, that's all what -state uses
17:04 < stepcut> Saizan: isn't Migrate in HAppS-Data ?
17:06 < Saizan> err, yeah, with Serialize i meant all the serialization part, with Migrate/Version/Serialize, but probably HAppS-Data defines some instances with generics for them too
17:07 < stepcut> I have used the serialize stuff independently of HAppS-State before. Not sure where I like it better.
17:08 < stepcut> there are reasons why either place could be right.
17:08 < Saizan> HAppS-Serialize then?:)
17:08 < stepcut> Saizan: I was thinking that...
17:10  * stepcut should be thinking about the test framework instead
17:13 < koeien> i could take a look at that. but probably that'll take too much time to be ready for the 0.1 release date
17:14 < stepcut> koeien: adding HAppS_Serialize?
17:14 < koeien> no, test-framework ;)
17:14 < stepcut> koeien: oh, I am doing that now. It's assigned to me ;)
17:14 < koeien> yeah, that's probably better
17:14 < koeien> it would take me too much time
17:15 < stepcut> I have everything mapped out, should be straight forward
17:15 < koeien> okay
17:42 < h_buildbot> Build for ghc-6.8.3 just succeeded. See http://buildbot.happstack.com/ghc-6.8.3/
17:44 < h_buildbot> Build for ghc-6.10.1 just succeeded. See http://buildbot.happstack.com/ghc-6.10.1/
17:44 < koeien> that's fast. what happened to the facebook module?
17:49 < wchogg> I thought mae was going to do something to tweak the build times.
17:50 < koeien> perhaps this morning (western europe time) only commits from you after that
17:50 < koeien> anyway, fine with me :)
18:35 < Lemmih> Hm, I wish people would realize that HAppS-DNS was no more included than HAppS-plugins.
18:35 < koeien> what is the point of HAppS-DNS btw? is there a use case for it?
18:36 < Lemmih> I have no idea if it even builds.
18:37 < stepcut> Lemmih: 77 minutes into this presentation alex j talks about the dream of disabling disk logging, http://www.bayfp.org/blog/2007/10/16/alex-jacobson-on-happs-videos-slides/
18:37 < Lemmih> Like with HAppS-Plugins, it was pretty much just a scrapyard for unused code.
18:37 < stepcut> Lemmih: sorry about getting you guys confused :)
18:38 < Lemmih> stepcut: It would just be an optimization to reduce latency.
18:40 < stepcut> Lemmih: practically speaking, disabling disk logging is probably not that useful. He said it more in the spirit of really big distributed clusters that were at least a megaton apart. Basically, the idea that you don't really expect all of googles servers to go down at once.
18:40 < Lemmih> There's this blog that complains about HAppS including in-memory state, SMTP, IRC and DNS. That's one right out of four. /-:
18:40 < stepcut> it was sort of an offhand comment, it wasn't a pre-planned pointed
18:41 < koeien> Lemmih: you mean http://aftnn.org/journal/661
18:41 < stepcut> Lemmih: I probably confused him with my happs based IRC bot :(
18:41 < Lemmih> koeien: Yes.
18:43 < koeien> h_buildbot is not HAppS-based though :)
18:43 < stepcut> Lemmih: some of us were talking about writing an alternative to HAppS-Server that is based around left-fold enumeration (possibly reusing hyena). Any idea if this has been discussed before?
18:44 < Lemmih> stepcut: I don't even know what that means.
18:44 < stepcut> http://www.galois.com/blog/2008/09/12/left-fold-enumerators-a-safe-expressive-and-efficient-io-interface-for-haskell/
18:46 < stepcut> Lemmih: late last year some people started doing web servers based around 'left-fold enumeration' instead of lazy IO. Supposedly it is faster, safer, and easier to reason about.
18:46 < stepcut> not really a short-term goal at this point in time
18:47 < stepcut> http://www.deinprogramm.de/defun-2008/abstracts/oleg-abstract.txt
18:48  * stepcut has not actually read the papers yet
18:48 < koeien> the presentation is very interesting. "why didn't i think of that?"
18:48 < stepcut> koeien: yeah, that's what I hear
18:49 < stepcut> I will move the topic to the top of my reading list
18:53 < doublec> stepcut, also http://www.bluishcoder.co.nz/2008/12/random-and-binary-io-using-iteratees.html
18:53 < doublec> that links to some other of Oleg's articles on the subject
18:53 < koeien> i think it's too early to determine whether it should be built in happs-server, but it's a cool idea
18:54 < stepcut> doublec: cool
18:55 < stepcut> koeien: yeah, I envision it as being an alternative for the extra cutting edge folks
18:55 < stepcut> koeien: ideally using hyena instead of writing a custom implementation
19:00 < koeien> why keep two alternatives? but again, i don't know the details so i don't have an opinion formed yet
19:00 < koeien> by the way, HAppS via SSL: https://huygens.functor.nl/happs-ssl/
19:01 < stepcut> koeien: nice! how?
19:01 < koeien> apache proxy.
19:02 < koeien> i'll write an article about it
19:02 < stepcut> on the wiki?
19:02 < stepcut> that definitely belongs in our core documentation somewhere
19:02 < koeien> i will document it
19:03 < stepcut> sweet!
19:03 < koeien> it's documented all over the web btw, but it's nice to have our own HAppS-specific docs
19:03 < koeien> rather, happstack* :)
19:03 < stepcut> right
19:04 < stepcut> most people (like me) probably had no idea to even look for that
19:04 < koeien> that's why it's good to have in the happstack docs
19:04 < stepcut> right
19:06 < stepcut> I wonder if there is a way to do it using s_server
19:06 < koeien> i can look into that
19:06 < stepcut> cool
19:06 < koeien> for my setup it was easier to insert an apache proxy, since there already was an apache on port 80
19:06 < koeien> (not on 443 btw)
19:06 < stepcut> right
19:07 < koeien> i don't have a valid cert for huygens, that's a pity :)
19:08 < stepcut> :)
19:43 < mae_work> hello
19:43 < mae_work> how is everything gonig?
19:43 < mae_work> going *
19:46 < koeien> https://huygens.functor.nl/happs-ssl/
19:47 < mae_work> neat
19:48 < koeien> did you tweak the facebook module?
19:48 < koeien> compile times are halved
19:49 < mae_work> koeien: nope
19:49 < mae_work> might have been your cleanup
19:49 < koeien> then it must be the cache or something like that
19:49 < mae_work> you made alot of those th errors go away right?
19:49 < mae_work> (wchogg)
19:49 < mae_work> rather wchogg did
19:49 < mae_work> I think that might have made it compile faster
19:51 < koeien> they're still present in the last ghc-6.10.1 build
19:55 < mae_work> hm
19:56 < mae_work> were they in the  6.8.3 build before?
19:56 < koeien> the buildbot logs seem to indicate that they were not
19:57 < mae_work> ok
19:57 < mae_work> could have been the load on your vps
19:57 < mae_work> was higher before
19:59 < koeien> maybe the load on other vps's on the server, or the disk cache is filled now, which makes ghc & darcs go faster
19:59 < mae_work> true
20:05 < mae_work> stepcut: how is test patch coming along?
20:06 < stepcut> mae_work: slowly. earlier today I didn't even have ghc installed on my machine :)
20:06 < mae_work> hehe
20:07 < stepcut> now I am trying to figure out how to read ghc's installed package list
20:07 < mae_work> Saizan_: get ahold of igloo regarding syb-with-class?
20:07 < stepcut> there is a function getInstalledPackages, but it requires lots of weird arguments
20:07 < stepcut> once I figure that out, I can fix cabal-debian, so that I can debianize happstack and install the damn thing
20:08 < stepcut> the rest is easy after that
20:08 < mae_work> hehe
20:08 < mae_work> partial application is your friend
20:09 < stepcut> I am looking at the source for ghc-pkg now
20:09 < stepcut> since ghc-pkg list some how figures it out
20:09 < mae_work> stepcut: i see :)
20:10 < stepcut> mae_work: do you use skype or anything other voip programs?
20:10 < mae_work> yes but i'm about to hit the road in a minute
20:10 < mae_work> you want to chat for a minute?
20:10 < stepcut> not now, but I am sure someday I will
20:11 < stepcut> sometimes typing just isn't good enough
20:11 < mae_work> stepcut: ok
20:11 < mae_work> stepcut: where do you live?
20:11 < stepcut> but, irc is good when it works, because it is easy for others to join in and it gets logged
20:11 < stepcut> chicago
20:11 < mae_work> ahh
20:11 < mae_work> my employer has a branch there
20:11 < mae_work> fresh logistics
20:11 < mae_work> in any case
20:12 < mae_work> i gotta run, i might be on later tonight.
20:12 < stepcut> my employee has a branch here now
20:12 < stepcut> employeer
22:35 < mae> hello
23:22 < stepcut> hello
23:22 < gwern> GREETINGS FELLOW HUMAN
23:23 < stepcut> mae: did you see any of the discussion about left-fold enumeration?
23:25  * stepcut goes to bed
23:25 < mae> stepcut: yes
23:25 < mae> stepcut: sounds nice, but I want to get this release out :)
23:26 < mae> gwern: hello thar
23:26 < gwern> mae: I did not greet you
23:41 < mae> gwern: my mistake :\
--- Log closed Thu Jan 29 00:00:11 2009