03:00:03 <rlpowell> stepcut: I really liked http://www.youtube.com/watch?v=7Wmszk4wZxQ
03:00:25 <rlpowell> Example-based documentation in the Haskell wordl is always so amazingly shitty/nonexistent; this was a refreshing change.
03:00:29 <rlpowell> I hope you're going to publicize it.
03:08:07 <donri> \o/
03:11:01 <rlpowell> ?
03:11:20 <rlpowell> (also, because I have 10 month old twins, it took me almost two weeks to watch the whole thing -___-)
03:11:23 <rlpowell> donri: Why you cheer?
03:11:51 <stepkut> :)
03:11:52 <donri> oh, just glad to see you here :)
03:12:09 <stepkut> I need to do the written documentation for happstack-foundation and then I can announce that video for real
03:12:26 <stepkut> thanks, btw!
03:12:27 <rlpowell> donri: Ah, OK.
03:13:12 <donri> maybe watching it over two weeks is good for retention
03:13:22 <stepkut> alas, I have been tied up working on a similar presentation for CUFP
03:13:49 <stepkut> that video was basically v2.0 of the presentation. and cufp will be version v3.0
03:14:13 <stepkut> hard to figure out how to condense it into < 45mins with out rushing through everything too fast
03:14:21 <stepkut> the first version was 3+ hours :)
03:14:50 <donri> the trick is to have a scaffold generator that is optimized for screencasts
03:17:37 <rlpowell> stepkut: Yeah, I don't know how you'd pare it down significantly.
03:17:47 <rlpowell> I would have rathered more explanation, not less.
03:27:35 <stepkut> rlpowell: right.. I could talk for 5 days.. easy :)
03:28:04 <stepkut> donri: well, with happstack-foundation.. there isn't really much boilerplate left
03:28:25 <stepkut> donri: for the remaining stuff, we need better lens, an auto-increment IxSet, and stuff like that I think
03:31:50 <rlpowell> I would *not* want it to have any boilerplate reduction from what you did.
03:32:10 <rlpowell> As a new Haskell user, there were already so many layers of abstraction as to make it hard to follow.
03:32:16 <rlpowell> Adding more in is too much, IMNSHO.
03:38:00 <donri> rlpowell: would you rather have more boilerplate, e.g. writing the TH stuff manually?
03:38:18 <donri> (that *is* a supported option for all our TH I think)
03:41:53 <rlpowell> donri: I don't think so, no, I think it's about right as it is.
03:42:00 <rlpowell> I'm just saying it's a crap ton to take in.
03:42:01 <donri> cool
03:42:05 <donri> in deed
03:42:15 <rlpowell> The TH stuff isn't where I get confused anyways, it's the 6 layer monads.  :)
03:42:44 <rlpowell> But with Happstack I feel like I have *some* idea what's going on and can alter it; Yesod was *too* deep.
03:42:53 <rlpowell> AFK, sorry.
03:47:10 <donri> 6 layer monads eh, let's not tell rlpowell about how serverpart itself is like 6 monads layered
03:50:26 <stepkut> :)
03:57:00 <luite> how does such a deep stack impact performance?
04:05:49 <stepkut> luite: unknown
04:07:07 <stepkut> luite: plan to experiment with it at some point though
05:30:12 <rlpowell> stepcut: Ah, I *can* suggest a place I'd like more boilerplate.
05:30:27 <rlpowell> (,,,) <$>  ... , that *entire* expression.
05:31:25 <rlpowell> Applicative is hard enough for a newbie; throwing in ++> and <++ is just 0___o
06:20:24 <rlpowell> What are people using for auth + sessions in happstack these days?
07:01:34 <rlpowell> -_-  Why does http://src.seereason.com/happstack-authenticate/demo-hsp/ require blaze?  :P
07:06:55 <rlpowell> I think it might just be broken.  *sigh*
08:07:29 <rlpowell> I got demo-hsp running!  \o/
08:07:35 <rlpowell> I had to make many changes.
08:07:44 <rlpowell> Is there somewhere I might send those to be helpful?
08:55:55 <Lemmih> Anyone with Windows care to help me test acid-state?
09:08:15 <Lemmih> Oh well. Here comes 0.7.1
11:51:31 <merijn> Lemmih: I have a question about acid-state implementation: Does it keep all data live in RAM? Or do you use something like mmap to have the OS load/page required data from disk on demand?
11:55:05 <srhb> merijn: Oh, was that the question. It keeps everythign in RAM, writing a history of all transformations on it to disk so it can be rewound after a shutdown
11:55:25 <merijn> srhb: There's "keeping data in RAM" and "keeping data in RAM"
11:56:19 <srhb> merijn: Hmm. Yes. And keeping data in RAM, lest we forget.
11:56:37 <merijn> srhb: The problem is that on a cheap VPS RAM is expensive, so having the entire data set in RAM might run you out of memory as your data set grows
11:56:49 <srhb> This problem is described on the homepage.
11:57:06 <srhb> To quote "Another potential solution is to create a special data-structure like IxSet which stores only the keys in RAM, and uses mmap and/or enumerators to transparently read the values from disk without loading them all into RAM at once."
11:57:15 <merijn> Whereas if you develop your code in such a way that the OS can page it out you might use a lot of address space without using a lot of actual RAM
11:57:24 <srhb> But just buy more RAM, or don't worry until you actually run out of RAM. I mean, how much are you planning to store?
11:57:40 <srhb> Even a Megabyte of data is a lot, for many use cases.
11:57:56 <merijn> srhb: Sure, I was just curious
11:58:10 <srhb> merijn: Have you read this? http://acid-state.seize.it/
11:58:47 <merijn> Yes
11:58:54 <merijn> Although that was a few weeks ago
11:59:12 <srhb> OK, any other details I guess I can't really help you with. :-)
15:18:43 <Lemmih> srhb: Thanks for being my proxy.
15:39:33 <srhb> Lemmih: I tried :-)
15:55:39 <Lemmih> srhb: You will be at ICFP, I take it?
15:55:51 <srhb> Too expensive!
15:56:04 <srhb> Lemmih: Sadly -- it sounds exciting!
16:37:09 <luite> same here :(