00:01:42 <donri> awake.until.you.watch.this.show.101.720p-dimension.r00
00:01:54 <donri> sounds promising!
00:02:09 <tazjin> Hm, I got the eztv release. Quality was perfect
12:52:00 <donri> stepcut: so, sessions. to monadstate or not to monadstate? it's not consistent with e.g. askRq ...
12:53:15 <donri> Eq can probably be dropped from my withSession by making it run a custom state type rather than simply State.
17:06:15 <Ngevd> I can't get sendFile from happstack-lite to work
17:08:18 <donri> in what way?
17:09:02 <Ngevd> For a start, the Haddock documentation lies
17:09:14 <Ngevd> Well, maybe not lies
17:09:54 <Ngevd> But it suggests using asContentType which gives IO String rather than ServerPart String
17:10:46 <donri> good catch, that does look wrong
17:11:31 <donri> the types are more generic in server proper and manually set to be more specific in lite. maybe stepcut never tested that one.
17:12:19 <Ngevd> Would I use... lift? liftIO? to get it to ServerPart
17:14:15 <donri> you could import the originals from Happstack.Server.FileServe
17:16:21 <Ngevd> Okay
17:16:23 <Ngevd> Thanks!
17:18:26 <Ngevd> Hang on, that wasn't the question I was stuck on...
17:18:35 <Ngevd> Is the Filepath it takes on the local system?
17:51:11 <Ngevd> So... is the FilePath operand in serveFile the file I want to serve's location on my hard disk?
20:58:50 <stepkut> oo, I see now! someone else complained about serveFile + asContentType as well.. but I couldn't figure out what they were getting at
20:59:19 <stepkut> fixed in darcs
21:00:19 <stepkut> the FilePath you give to serveFile is the location of the file on your disk
21:01:04 <stepkut> there are no safety checks done .. so if you are getting the filename from the user, you need to check that they are not doing something evil like ../../etc/passwd
21:01:23 <stepkut> we just added new functions to help with that, but they are only in darcs at the moment
21:10:10 <stepkut> i want to port acme-http to agda.. just to see how fast/slow it is
22:49:22 <donri> stepcut: how essential is it that we do the tricks to enforce the CookieLife of the clientsession? can it wait to a later release?
22:49:51 <stepkut> sure
22:50:20 <hpaste_> donri pasted should be usable now at http://hpaste.org/67181
22:51:41 <stepkut> neat
22:53:19 <donri> i'll add it in darcs
22:53:54 <stepkut> sweet
22:57:43 <donri> http://darcsden.com/dag/happstack/browse/happstack-clientsession \o/
22:59:21 <stepkut> \o/
23:13:48 <donri> stepcut: nibro released new src-exts that fixes hsx in do notation
23:15:20 <stepkut> donri: yup
23:15:35 <stepkut> donri: we are working on a bunch of other things too
23:16:12 <stepkut> donri: for example, in the darcs version, trhsx adds a type signature to all the String literals it generates so that you can enable OverloadedStrings
23:16:19 <donri> nice
23:16:26 <donri> did he pull your stuff for qq?
23:16:31 <stepkut> yes
23:16:38 <stepkut> we are also going to refactor things
23:16:48 <donri> christmas is early!
23:16:58 <stepkut> right now it looks like the HSP package will go away
23:17:25 <stepkut> the hsx package contains two separate things -- the XMLGenerator class, and the HSX.Transform / trhsx stuff
23:17:57 <stepkut> so, we are looking to split HSX.XMLGenerator into a separate package that includes the XML type, the renderAs* functions, and the XMLGenerator Identity instance
23:18:14 <stepkut> and nothing will depend on HJScript/HJavascript
23:18:54 <stepkut> I also proposed that we do not use Undecideable or Overlapping instances in the core
23:19:23 <stepkut> something like this instead: http://src.seereason.com/xmlgen-demo/
23:19:42 <stepkut> so that you do not get those horrible page long overlapping instances errors for EmbedAsChild
23:19:56 <stepkut> not sure if it is viable.. but the proof of concept attempt works
23:20:04 <stepkut> if only I could get rid of XMLGenT :(
23:20:38 <stepkut> now I am upgrading happstack-authenticate to the latest authenticate.. which means I had to send a patch to fb to upgrade that to the latest conduits :(
23:27:21 <donri> if you can get rid of the overlapping instances without losing a lot that'd be awesomesauce
23:28:05 <donri> there is of course some logic to that insanity but it's very cryptic. slightly change something that was working, or add something that seems to be just as the existing working code and it explodes in your face. :P
23:28:14 <stepkut> donri: yup
23:28:22 <stepkut> I think that is the #1 usability issue with hsx
23:28:29 <donri> yea
23:29:18 <donri> why do you think xmlgent is so bad? how is that different from any other need for a transformer?
23:31:33 <stepkut> is there someway to download this change as a .diff ? https://github.com/stepcut/fb/commit/2356b682ee7b464c7cd25f3103e0b025ae9c2b66
23:31:41 <stepkut> aka, .patch
23:34:47 <donri> yea, add .patch to the url
23:39:13 <stepkut> sweet
23:48:02 <stepcut> so tomorrow I will look at releasing updated happstack-server, hsx-jmacro, and happstack-clientsession libraries
23:48:14 <stepcut> and if the fb patch gets pull, happstack-authenticate
23:49:08 <tazjin> Do you have a changelog already?
23:49:20 <tazjin> I need something relaxing to read, this cookbook is killing me
23:49:59 <stepcut> tazjin: for what?
23:50:08 <tazjin> All the things
23:51:02 <stepcut> tazjin: the raw changelog is here, http://patch-tag.com/r/mae/happstack/snapshots/all/history
23:51:23 <tazjin> Well, thanks. That is more relaxing than writing a cookbook
23:51:32 <stepcut> tazjin: :)
23:52:12 <stepcut> tazjin: nothing earth shattering, just some nice small changes that there is no reason to sit on
23:52:32 <stepcut> the earth shattering development is happening elsewhere at the moment
23:52:36 <stepcut> ;)
23:52:42 <tazjin> In a secret laboratory
23:52:51 <tazjin> in the middle of a deserted forest, a high security facility
23:53:12 <stepcut> :)
23:56:03 <stepcut> ACTION considers dsfox's idea to rename digestive-functors 0.2 to formlets with increasing seriousness