17:22:22 <Gracenotes> hm.. if initializing the tx control takes a long time, is there a way to be able to shut it down while it's loading in a non-dangerous way?
17:22:49 <Gracenotes> you can add a signal handler after you obtain it, but not before. not sure recursive do would help here either
17:48:24 <stepkut> I am not really clear what the issue is here.. if you kill the process will it is still loading the state, nothing bad should happen
17:49:19 <stepkut> be back later, have to entertain a 3 year old before he destroys the place
21:52:36 <Lycurgus> Gracenotes, ping
21:52:53 <Gracenotes> ah pong
21:53:23 <Lycurgus> what's your assessment of happstack for practical, production use?
21:55:55 <Lycurgus> (assumming conventional RAM provisioning in the 2-4 GB range)
21:56:57 <Lycurgus> and in the right Laconic style you may certainly answer by silence
21:57:45 <Gracenotes> I think the laconics gave short pithy answers, rather than sat there a bit stumped
21:58:12 <Gracenotes> well, I've only run hackage on localhost and not as a public website, so my impressions are necessarily incomplete. however, I've done a lot of pushing code around and writing new code, and for SimpleHTTP, it's been easy and flexible to interface with it
21:58:34 <Gracenotes> the combinators it provides are good, and you can build your own abstractions onto them
21:59:37 <mightybyte> I'm running a production website on happstack
21:59:55 <Gracenotes> the combinators also make things very open-ended, so I'm not sure how things are going on the middleware front; it's easy to implement some HTTP features, but perhaps said features should be in the server itself (and many are on the roadmap)
22:00:18 <Lycurgus> mightybyte, URL?
22:00:49 <Gracenotes> happstack-state is nice (just keep your boilerplate-ish state declaration code off in a separate module), although I'm not not too familiar with migrations, and I'm not sure if there should be a better approach to it
22:01:02 <mightybyte> My server process has pretty good uptime.  It's been up about two weeks right now and before my last restart it had been up over a month.
22:02:11 <Lycurgus> "production" doesn't necessarily mean Bank of America or Google, but it also doesn't mean your science project
22:02:47 <Gracenotes> I don't think anyone will ever be as production-y as google
22:03:08 <Lycurgus> at a minimum it means carry a load of a few tens of thousands of visitors a day and in peak times thousands at a time
22:03:42 <Lycurgus> and of course doing stuff other than serving static content
22:04:32 <Lycurgus> Gracenotes, yes someone will be.
22:05:05 <Lycurgus> *something
22:07:34 <Gracenotes> performance, on multithreaded/multicore particularly, is pretty important to the happstack team. I think the MonadPlus approach is nifty for dynamic URIs, too
22:08:47 <Gracenotes> there have been benchmarks done, though of course I don't have them on hand... I think attoparsec is on the to-do list for HTTP pasing
22:08:47 <Gracenotes> parsing
22:13:53 <Lycurgus> i see. mightybyte's site appears to perform well.