10:49:31 <jonkri> When routing, how can I get the remaining directories (not just the next one, like with `path')?
12:16:53 <jonkri> I want to generate a HTML string using HSP, but when I include the string in a <% %> block then it's automatically escaped. How can I get around this? Do I have to generate my HTML using XMLGenT?
13:07:03 <stepkut> jonkri: <% CDATA False <lazy text value> %>
13:07:54 <stepkut> <% CDATA False (htmlString :: Data.Text.Lazy.Text) %>
13:14:10 <jonkri> stepcut: Perhaps the documentation should mention what that boolean means (parsed vs unparsed?).
13:15:25 <jonkri> stepcut: Anyway, when I do "[hsx| <div<% if b then return $ CDATA False "style=\"margin-right: 0px;\"" else return $ CDATA False "" %>></div> |]" in a "XMLGenT (HSPT XML Identity) XML", I get an error "Couldn't match type `XMLGenT m0 [ChildType m0]' with `[a0]'; Expected type: [a0]; Actual type: GenChildList m0". What does that mean?
13:16:12 <Clint> why are you return-ing?
13:16:38 <stepkut> oh, you want to use the value as an attribute value not a child ?
13:16:40 <jonkri> I thought I had to go from "XML" to "XMLGenT (HSPT XML Identity) XML". Let me try without.
13:17:00 <jonkri> stepcut: Ah, yes, of course there's a difference there. Sorry. :-P
13:17:26 <stepkut> CDATA can only be used for children
13:17:37 <jonkri> That makes sense. :-)
13:18:55 <jonkri> What can I do instead?
13:19:15 <stepkut> you have a pre-escaped attribute value?
13:19:50 <jonkri> stepcut: Yes. "margin-right: 0px;".
13:21:02 <stepkut> to do what you wrote there I would just do, <div (if b then [("style" := "margin-right: 0px;")] else []) ></div>
13:21:47 <stepkut> or you can do: <div [("style" := "margin-right: 0px;") | b ]></div>
13:22:03 <stepkut> if you want to make clever use of list comprehensions
13:25:21 <jonkri> Aha...
13:25:56 <jonkri> No instance for (EmbedAsAttr (HSPT XML Identity) (Attr n0 a0))
13:25:56 <jonkri>       arising from a use of `asAttr'
13:25:56 <jonkri>     The type variables `n0', `a0' are ambiguous
13:31:18 <stepkut> you have OverloadedStrings enabled?
13:31:39 <jonkri> Yes.
13:32:11 <stepkut> if so, you'll need an explicit type sig like, [[("style" := "margin-right: 0px;") :: Attr TL.Text TL.Text | b ]
13:32:15 <stepkut> which is really annoying
13:32:27 <jonkri> If I remove it, I get:
13:32:29 <jonkri>     No instance for (EmbedAsAttr
13:32:29 <jonkri>                        (HSPT XML Identity) (Attr [Char] [Char]))
13:32:29 <jonkri>       arising from a use of `asAttr'
13:33:54 <stepkut> yeah.. for some reason we did not provide an EmbedAsAttr instance for (Attr String String)
13:34:00 <jonkri> With the type and OverloadedStrings enabled, it works. Thanks a lot! :-)
17:55:09 <donri> ACTION writes an immensely complex patch to port his code to the new hsx2hs
17:55:15 <donri> hint: it consists of fromStringLit = id
18:02:29 <sm> bah donri
18:07:45 <donri> sm: in fairness i'm not even using hsp here ;)
18:07:54 <donri> it's not even for html...
18:26:35 <donri> stepkut: say do we care about IE8 and below? :)
18:26:46 <donri> apparently IE9+ supports XHTML5
19:29:22 <kxra> donri: http://normanlayout.info/compare
19:29:24 <kxra> ahhhh
19:41:30 <stepkut> donri: yeah, I have that defined in some code as well (fromStringLit = id)
19:41:33 <stepkut> 'tis perfectly valid
20:19:16 <donri> kxra: i saw norman, but wasn't convinced by his arguments for it and didn't investigate much further
20:19:34 <donri> maybe i should
20:21:41 <stepkut> I'm probably going to stick with dvorak
20:21:47 <stepkut> widely available and I know it already
20:21:59 <stepkut> not sure the return is there on learning yet another layout
20:22:18 <donri> i'm building a hardware programmable keyboard which nullified one of my reasons for colemak of "availability"
20:23:39 <stepkut> :-/
20:23:56 <donri> norman seemed interesting because it addressed some of my pain points from colemak, is optimized for matrix keyboards and his "reach difficulty" scoring model made sense to me
20:24:02 <donri> uh, workman*
20:24:29 <stepkut> I'm not really clear how a programmable keyboard helps with availability
20:24:38 <donri> stepkut: i can hardwire the layout
20:24:46 <stepkut> yeah.. but then you have to take your keyboard everywhere
20:24:54 <donri> yes, i wouldn't want any other keyboard anyway ;)
20:24:57 <stepkut> i like dvorak because it comes pre-installed on linux, os x, and windows machines
20:25:04 <donri> yeah
20:25:21 <stepkut> so I can easily switch to it when using a machine that is not my own
20:25:32 <donri> each their own and all that. i almost never touch other computers, and when i do i just stick to qwerty
20:25:39 <stepkut> a hardware keyboard would do that as well.. but i'd have to take it with me :)
20:26:10 <stepkut> I also find per-window layouts to be useful for playing nethack ;)
20:26:18 <donri> still i find it a bit cumbersome to have to set up colemak everywhere... linux tty is separate from x11... gnome is separate from x11... each VM is separate...
20:26:26 <donri> even though colemak is there, i have to set it up
20:27:21 <donri> the worst part is that i can't type my passwords in qwerty, so i can't log in on tty xD
20:30:34 <donri> kxra: i almost want to use a key logger for a few months and generate the optimal layout from that xD
20:33:21 <donri> you'd probably need proper NLP tools do to it right
20:33:32 <donri> to account for bigrams etc
20:34:41 <kxra> donri: surely it has been done?
20:34:48 <donri> stepkut: even better this keyboard is a split design of two heavy pieces one for each hand... so easy to take with you! ;)
20:34:52 <kxra> or do you mean one personalized for you
20:35:04 <donri> kxra: yes, personalized based on past key presses
20:35:21 <donri> kxra: carpalx etc use public domain texts
20:35:43 <donri> ie. 19th century english literature :p
20:36:37 <donri> also they don't take the NLP idea to the extreme AFAIK
20:39:02 <donri> let's write a haskell program to generate layouts based on parameters and corpus!
20:41:12 <donri> stepkut: isn't it great BTW how the best layout for touch typing is the worst for touch screen typing aka "swyping" :D
20:41:48 <donri> for swyping you basically want the worst possible layout to minimize ambiguity
20:47:41 <kxra> donri: that sounds awesome
20:48:00 <kxra> donri: yeah that makes sense
20:48:07 <kxra> about swyping
20:48:38 <donri> kxra: actually i think this is what carpalx is, but in perl and without the parameters used for workman
20:54:14 <donri> kxra, stepkut: what keyboards do you use?
21:52:12 <stepkut> donri: keyboards? or layouts?
21:52:28 <donri> stepkut: keyboards
21:53:14 <stepkut> the one built into the T61 which is cursed by the most ill-designed touchpad on the planet
21:53:24 <stepkut> and the standard external apple usb keyboard
21:53:48 <stepkut> I have a happy hacker keyboard lite II and an IBM Model-M around, but not in use
21:54:07 <donri> the lite model is rubber dome right?
21:56:14 <stepkut> no idea
21:56:41 <stepkut> it's like this, http://www.tablix.org/~avian/blog/images2/2009/07/happy_hacking_keyboard_lite2_usb.jpg
21:57:21 <stepkut> the ibm is nice because it goes 'sproing' every time you type a key and you can easily move the key caps around
21:57:46 <stepkut> except if you move the F and J keys then it is hard to tell if you are properly centered on the keyboard
21:58:19 <stepkut> time to bevel some edges!
21:59:13 <donri> even so the key caps are likely sculpted such that you can't rearrange them column wise
21:59:25 <donri> well, you can but it would be weird
22:04:29 <donri> stepkut: if you like buckling springs there's the unicomp that you can order with dvorak key caps and even a thinkpad style pointing stick
22:04:29 <stepkut> on the ibm model-m almost any keycap can be swapped except for modifier keys
22:05:02 <donri> stepkut: sure, but i'm not sure the shapes will be identical
22:05:16 <stepkut> I mostly liked to have a stupidly loud keyboard in an open office environment -- now that I work at home it doesn't matter :)
22:06:42 <donri> i have a clicky mechanical switch keyboard but buckling springs are supposed to be louder or at least thicker sounding
22:10:29 <donri> stepkut: most keyboards are 'contoured' http://deskthority.net/w/images/5/5b/KeyProfile.gif
22:10:33 <donri> not sure about model m
22:10:56 <donri> so you can move the key caps around but it will result in an uneven "profile"
22:22:59 <donri> bedtime