03:07:32 <stepkut> bah. too much boilerplate. time to do something about it :)
07:24:11 <levi> So, I am going through the happstack crash-course and I also looked at clkwrks a bit.  I'm curious why clkwrks uses markdown.pl instead of something like pandoc.
07:26:33 <donri> maybe licensing?
07:30:58 <levi> Ahh, I guess pandoc is GPL.  What's clkwrks license?
07:31:26 <donri> probably BSD
07:31:55 <donri> it could use the sundown package though instead
07:34:28 <levi> Yeah, I think there are a bunch of C Markdown libraries, I just thought it would be nice to use a native haskell one and I've used pandoc before.
07:36:42 <levi> pandoc is especially nifty due to the multitude of input and output types, built-in template system, etc.
07:36:56 <donri> yea, pandoc is great
07:41:54 <levi> I wrote a Markdown parser for Common Lisp a few years ago.  Fortunately, it appears to have disappeared from Google now. :)
07:43:06 <donri> of course, markdown itself sucks
07:43:59 <levi> I like it's 'user interface' so to speak, but not its linguistic properties.
07:44:07 <levi> its
07:45:55 <donri> it's lacking standardization, extensibility, kind of crappy for writing in textareas
07:46:20 <donri> usually when people say "markdown" they mean "our own custom bastardization of something resembling markdown"
07:47:21 <levi> Yeah, but that seemed to kind of be the point of Markdown in the first place.  You just write nice-looking ascii following some conventions and it automatically converts to html.
07:47:55 <donri> well, the cake is a lie
07:48:57 <donri> i like reST and creole, although they're not well supported in haskell :(
07:49:17 <levi> Pretty much everything else is a lot more markup-heavy and more awkward to write in, even if more standardized.
07:50:16 <levi> reST is fairly nice, though.
07:50:31 <donri> duno, markdown can be pretty when you edit it in a proper editor and you're doing something insanely simplistic
07:50:53 <donri> but, try writing a link that includes parenthesis
07:51:08 <donri> try staying inside a column margin like 80 characters
07:51:34 <donri> try writing a large code block in a textarea web form
07:52:13 <levi> I would probably become annoyed at Markdown if I had to do any of those things with any frequency.
07:52:27 <donri> yea, that's like all the time for me :)
07:52:42 <donri> i end up opening vim just to comment on reddit
07:53:15 <donri> and even then no vimscript gets the syntax highlighting right
07:53:49 <donri> especially since, again, custom bastardizations so markdown written for github ends up looking shit in vim
07:55:02 <donri> and people get it wrong all the time on reddit making their comments less readable due to messed up formatting
07:55:03 <donri> ^_^
07:55:38 <levi> Well, people are going to get any kind of markup wrong all the time on a site like reddit.
07:56:08 <donri> well, the problem is there are awkward edge cases in markdown, or maybe it's just reddit's markdown, again, who knows what markdown even is
07:57:44 <donri> like you have to end a line *with a space* if you want to break a paragraph
07:59:46 <levi> Yeah, that bit is rather funky.  I shall not attempt to further defend Markdown. :)
08:00:06 <donri> haha
08:00:12 <donri> /rant ;P
08:01:56 <donri> oh and then people mix markdown and yaml (e.g. jekyll) because markdown has no concept of document meta data
08:02:06 <donri> which also looks horrible in vim :)
08:02:18 <donri> ACTION couldn't resist ranting a bit more
08:03:09 <donri> and they don't even do it in such a way that it's a valid yaml document ^_^
08:03:58 <levi> When your document needs metadata, you should probably move on from markdown.
08:04:38 <donri> yea, but people don't, because not-quite-markdown is everywhere
08:04:47 <donri> it's an addiction i say!
08:05:10 <levi> A coworker of mine is working on an IEEE standard that the editor has chosen to use reST for.
08:05:19 <donri> we should just use haskell combinators instead
08:06:09 <donri> the irony of reST is that docutils is very badly documented ^_^
08:06:53 <levi> It appears to be better documented than Markdown, at least!
08:07:21 <donri> duno about that, it's crazy more powerful
08:09:14 <levi> We should just write directly in Docbook XML.  Or DITA XML. :P
08:12:12 <donri> asciidoc!
08:12:30 <levi> No, no... you will write out all your tags by hand!
08:13:23 <donri> with blaze-markup right
08:13:53 <donri> wouldn't it be great if you had to know haskell to comment on reddit
08:14:33 <levi> The reddit from an alternate timeline where they initially wrote it in haskell instead of lisp, and then didn't have to rewrite it.
08:15:22 <donri> and then they rewrote it in agda
08:16:23 <levi> I actually wrote a software design document in docbook as an experiment.  Writing it wasn't too bad, but editing and restructuring parts of it was really painful.
08:17:28 <levi> The rendering toolchain went through latex, so the output was really nice. :)
08:18:57 <levi> I also found some ascii-based diagram tools that let me embed the source for the diagrams in the document.  We still use those for doing sequence charts.
08:22:27 <donri> yea graphviz
08:25:30 <levi> Nah, this was called PlantUML or something.
08:26:16 <donri> aha
08:33:00 <levi> I played a bit with the haskell diagrams library more recently.  I wanted to visualize the result of an algorithm I was working with and managed to get a pretty nice picture out of it.  It's got an SVG backend, so it could be interesting to hook into a web app.
08:37:21 <levi> One final thing before I go to bed.  This site has an interesting concept for CMS-style stuff: http://interior.substance.io/
08:38:58 <levi> It's heavily javascript-oriented at the moment, but I think the model would lend itself well to a haskell implementation
13:16:10 <stepcut> levi: clckwrks needs a custom markdown-like parser anyway so that plugins can easily add extensions and stuff
13:28:13 <donri> stepcut: btw pt sans is gorgeous with autohinting ;)
13:28:29 <stepcut> donri: >:(
13:28:31 <donri> :D
13:28:40 <stepcut> :P
13:28:41 <donri> been playing with fontconfig
13:28:59 <stepcut> nice
13:29:32 <donri> i personally like non-auto hinting but it only looks nice for a few carefully hinted fonts
13:29:38 <stepcut> I'm working on a little library that makes it easier to get started with happstack+hsp+web-routes.. i hope
13:30:12 <donri> and it jumps from thin lines to fat ones with no steps as the point size grows
13:30:21 <donri> aha sounds interesting
13:31:10 <stepcut> yeah.. should be pretty straightforward but useful
13:31:24 <donri> also i found myself rewriting a top-level pattern matching function similar to route functions, and separating it into separate functions like i said didn't seem useful to me before
13:31:30 <stepcut> get rid of some boilerlpate
13:31:32 <stepcut> plate
13:31:40 <stepcut> haha
13:31:50 <donri> because using the same function is less nice with my folds in vim
13:32:05 <stepcut> :)
13:32:06 <donri> so maybe your route function TH would be useful to me too
13:32:23 <stepcut> :)
13:32:35 <donri> i wonder if it could be generalized beyond web-routes though
13:32:46 <stepcut> no idea
13:33:55 <donri> bbl
13:34:43 <stepcut> k
14:15:49 <Igloo> Is there any lockfile code in happs anywhere?
14:15:56 <Igloo> Or in a library that happs depends on?
14:18:04 <donri> all i know is stepcut once said portable file locking is hard
14:19:23 <stepcut> Igloo: no.. there was some in happstack-state, but it never worked quite right and did not get into acid-state
14:20:15 <Igloo> Does anything stop you from running 2 instances of a happs server, then?
14:21:26 <stepcut> Igloo: Not in happstack itself. Though, if we could get a working portable solution that would be nice. There are some open bugs..
14:21:50 <Igloo> OK, ta
14:22:30 <donri> if you use the same port you'll get an error
14:24:02 <donri> Igloo: http://hackage.haskell.org/packages/archive/unix/2.5.1.1/doc/html/System-Posix-IO.html#g:7 if you're only on posix
14:24:57 <Igloo> ta
14:25:13 <Igloo> I'm first looking to see whether I can abuse start-stop-daemon
14:25:40 <Igloo> (I'm not actually trying to use it for a happs server itself, but for the hackage doc-builder)
14:30:09 <donri> http://hackage.haskell.org/packages/archive/posix-filelock/0.1/doc/html/System-Posix-FileLock.html looks nice as well