--- Log opened Mon Apr 06 00:00:08 2009
00:13 < performance> @seen sm
02:19 < vegai> happs has some form of concurrency issues
02:19 < vegai> happstack-0.2.1 that is
02:19 < vegai> I'm able to kill the server with ab -k -n32768 -c4
02:19 < vegai> where -c2 and -c1 go through just fine
02:20 < vegai> by kill I mean it goes into a state where it drains >100% CPU and doesn't serve any requests anymore
11:04 < stepcut> I am going to commit this fix for 'cron' unless someone sees a flaw, http://hpaste.org/fastcgi/hpaste.fcgi/view?id=3415#a3415
11:23 < h_buildbot> Build for ghc-6.8.3 OK. Test suite ran. 14/77 test cases failed. Check http://buildbot.happstack.com/ for details.
11:29 < h_buildbot> Build for ghc-6.10.1 OK. Test suite ran. 16/77 test cases failed. Check http://buildbot.happstack.com/ for details.
11:55 < stepcut> dons: you around?
12:21 < dons> yo
12:52 < stepcut> dons: I looked at trying to use hint or the ghc-api as a replacement for plugins, but I don't see all the pieces. the ghc-api definitely has better code for compiling stuff. But I don't see a way to load / link symbols into the running process the way plugins did. Do you know if I am missing something obvious? Or does that functionality need to be ported from plugins?
16:39 < lanaer> recently updated to ghc 6.10.2... and now I get "Unsupported socket" on every connection attempt, whether it be to the sample guestbook app or to my own apps :(
16:39 < lanaer> anyone else seen a similar problem? (Intel Mac OS X 10.5)
21:21 < mae> ladies and gents
21:21 < mae> hello
21:41 < stepcut> mae: I fixed a bug today :)
21:41 < stepcut>  
21:41 < stepcut> well, it was a joint effort
21:44  * gwern waits for the next beat. well, it was really a doc bug. well, it really was just stylistic. well, I *thought* about changing it.
21:58  * stepcut has also thought about fixing some doc bugs
22:00 < stepcut> the docs for FromReqURI are wrong IMO
22:05 < mightybyte> http://softwaresimply.blogspot.com/2009/04/standalone-auth-framework-for-happstack.html
22:06 < mightybyte> More to come in the next few days.
22:07 < stepcut> mightybyte: this is neither valid html nor valid xhtml, http://github.com/mightybyte/happstack-auth/blob/3b9149605e819b7c4657e5905d4f4ad94b29404b/demo/login.html
22:10 < mightybyte> It wasn't supposed to b.
22:10 < mightybyte> be
22:10 < mightybyte> Oops, I did miss <head>...that wasn't supposed to happen.
22:10 < stepcut> among other things...
22:11 < mightybyte> What other things?
22:12 < stepcut> some <input> fields are short tags <input ... /> which implies xhtml. But, some are not -- which means xhtml would throw an error if they actually tried to parse it.
22:12 < mightybyte> Ok, fixed.  Anything else?
22:13 < stepcut> the input field is not actually supposed to be instead the label field. Not sure if that is technically invalid or not.
22:13 < mightybyte> s/instead/inside/ ?
22:14 < stepcut> oops, sorry been drinking :) I did mean inside.
22:14 < mightybyte> Changes pushed.  Thanks.
22:17 < stepcut> if you intend this to be xhtml that will be parsed as html, then it is recommended that you put a space before />
22:17 < stepcut> http://www.w3.org/TR/xhtml1/#C_2
22:18 < stepcut> however, if you intend for this to be xhtml, then you should really validate it, because xhtml parsers are required to throw an error if anything is wrong
22:18 < stepcut> if you mean for it to be html, then html does not support shorttags
22:19 < stepcut> also, textfield is not a valid type for <input type=... >
22:20 < mightybyte> What is it...just "text"?
22:20 < stepcut> yes
22:21 < stepcut> also, you can't just stick label or input inside a form element
22:21 < stepcut> the only valid children of a form element are, P, PRE, ADDRESS, DIV, and H1-H6
22:23 < mightybyte> Ugh
22:23 < mightybyte> Standards schmandards
22:23 < stepcut> :)
22:24 < stepcut> there is no point in writing invalid xhtml. If you don't want to validate, at least try to write plain old html :)
22:24 < mightybyte> Heh, the problem is that I don't know the difference.
22:24 < stepcut> if you installed wdg-html-validator and use validateConf instead of nullConf, then happstack will validate your pages on-the-fly
22:25 < mightybyte> There, I fixed a few more.  The html really isn't the focus, so I probably won't spend time on it.
22:25 < mightybyte> Ahhh, cool.
22:26 < mightybyte> If you want to fork it in github and fix those issues, I'll pull them in.
22:26 < stepcut> also, if you username is <script type="text/javascript">alert("foo");</script>, fun stuff may happen
22:27 < stepcut> or if you username contains unicode, it may not render correctly
22:27 < dons> congrat to the tupil team for finishing their first commercial (happstack !) web system
22:27 < dons> http://twitter.com/tupil <-- did you guys know about this?
22:28 < mightybyte> Nope
22:28 < mightybyte> That's great.
22:28 < dons> might want to add it to the "2 new projects"
22:28 < dons>  "Update: *3* new projects!"
22:28 < dons> http://blog.happstack.com/2009/04/06/2-new-projects-that-use-happstack/
22:29 < stepcut> dons: is that the formlets guy?
22:29 < stepcut> dons: chris e. ?
22:29 < mightybyte> Yeah, one of them.
22:29 < dons> it's eelco_ and chris
22:29 < dons> does happstack.com list the known projects? gitit , patch-tag et al?
22:30 < dons> it'd be cool to have a nice list to refer people to
22:30 < stepcut> dons: I looked into replacing hsplugins with ghc-api. ghc-api has some nice stuff for recompiling modules. But I did not see a way to load the new symbols into the running executable the way that hsplugins did...
22:30 < dons> yeah, the lower level api would still be needed
22:30 < mightybyte> Well, I'm off.  Good night.
22:30 < stepcut> dons: ok, just checking. What part was broken by 6.10?
22:33 < dons> not sure.
22:34 < stepcut> ok
--- Log closed Tue Apr 07 00:00:08 2009