Experimental IRC log happs-2008-01-09

Available formats: content-negotiated html turtle (see SIOC for the vocabulary)

Back to channel and daily index: content-negotiated html turtle

These logs are provided as an experiment in indexing discussions using IRCHub.py, Irc2RDF.hs, and SIOC.

06:22:39<vegai>alexj_: typo on the new page, "Is provides event subscription" => "It provides event subscription"
06:22:48<vegai>unless you were going for lolcat style or something :P
07:50:30<vegai>hmm, you changed the page away again?
08:46:26<mmmdonuts>I'm getting an error building src/HAppS/Store/HelpReqs.hs in the HAppS-HTTP repo:
08:46:51<mmmdonuts>src/HAppS/Store/HelpReqs.hs:97:30:
08:46:53<mmmdonuts> No instance for (Default GetHelpReqs)
08:46:55<mmmdonuts> arising from a use of `webQuery'
08:46:56<mmmdonuts> at src/HAppS/Store/HelpReqs.hs:97:30-37
08:46:58<mmmdonuts> Possible fix: add an instance declaration for (Default GetHelpReqs)
08:46:59<mmmdonuts> In the first argument of `($)', namely `webQuery'
08:47:01<mmmdonuts> In a 'do' expression: helpReqs <- webQuery $ GetHelpReqs
08:48:06<mmmdonuts>This happens with the latest version, as well as with tag 0.9.1.2. Also happens in the old HAppS-Server repo. Which makes me think I must be doing something wrong...
08:50:17<mmmdonuts>ACTION notices that this is to provide help via the web interface, which I don't need
10:33:36<mmmdonuts>HAppS seems to be in really good shape right now. I'm impressed.
10:35:22<Sizur>did the morphisms do their job already?
10:35:44<Sizur>i mean the api is stable enough to start developing on it?
10:37:04<mmmdonuts>Well, it's mainly that I was able to build three separate apps that I had done at different times into a single happs instance without any trouble.
10:37:29<Sizur>btw guys, i have done an early version of i18n, it's in hackage. doesn't yet support plural forms. so if anybody wants to use it, be my guest and please email me with comments
10:38:54<Sizur>is a template engine badly needed? if we will implement a TemplateToolkit-like engine, will it be welcome?
10:41:33<mmmdonuts>Personally, I'd like to be able to use something like HSP with happs, where you can embed HTML directly in Haskell code.
10:42:08<mmmdonuts>But it seems as if everyone has a different opinion about this (and I wouldn't use that approach for everything, either).
10:44:01<mmmdonuts>Simple template engines tend to be limiting.
10:44:34<mmmdonuts>You need a good solution for reusing templates within other templates (better than just textual inclusion).
10:46:04<mmmdonuts>Ideally, you want to be able to invoke one template from another and pass it parameters. Otherwise, you end up having to fake something like that anyway.
10:46:17<Sizur>TemplateToolkit is the best i know of
10:54:55<mmmdonuts>I haven't used it. Looking at the docs, it looks like it handles some of the things I mentioned. If a good job was done of integration with Haskell features, something like that could be useful.
10:58:16<Sizur>i don't know what you cannot do with TT, but many things will have to be cut from it to be used for haskell
11:00:28<mmmdonuts>I'd be a bit concerned about ending up with the worst of both worlds - you'd want to be able to use maps and folds in your templates instead of for loops, for example.
11:01:34<mmmdonuts>I think it'd be better to allow embedding Haskell code than supporting a distinct template programming language, otherwise the template system just becomes a barrier.
11:02:59<Sizur>i thought the same, until #haskell bashed more sense into me
11:03:28<Sizur>i wanted a combinator lib for templating
11:03:41<Sizur>but i do see that it will not be convinient
11:05:50<mmmdonuts>What sense did #haskell bash into you? Bash me! :)
11:09:33<Sizur>simply that it will be counter productive to have a combinator lib for templating as it will not be convinient to use/read
11:09:50<Sizur>if it's not convinient, coder will not use it
11:12:15<mmmdonuts>Arguable, the XHTML lib is a combinator lib for templating. But perhaps it proves the point.
11:12:49<Sizur>yeah, just for a specific, limited output
11:13:55<Sizur>one can argue that normal list functions already give us the templating engine. but it's slow
11:18:00<mmmdonuts>Yes, I looked at Oleg's HSXML: http://okmij.org/ftp/Scheme/xml.html#typed-SXML
11:19:01<mmmdonuts>...which uses lists to represent XML (again, not general templating, but the idea could be used for other things).
11:19:50<mmmdonuts>But the implementation wasn't complete and the type errors it produces when a mistake is made are truly epic.
11:22:16<Sizur>sxml looks nice
11:23:41<mmmdonuts>SXML in Scheme is great, I've used it quite a bit. Embedded in Haskell, I think it needs some work. Actually a template engine which understood that syntax (as opposed to relying on embedding in Haskell code) could be useful.
11:26:46<Sizur>actually i like a lot how sxml looks in haskell
11:30:04<mmmdonuts>I like the look of it too. It was the type errors and the incompleteness of the implementation that put me off.
11:33:17<Sizur>what's missing in the impl?
11:35:41<mmmdonuts>It seemed like only a core set of HTML elements were implemented. Adding new ones didn't seem that easy, since there's heavy use of typeclasses etc. I didn't seriously try to extend it.
11:36:00<mmmdonuts>I thought I might come back to it later, and that was just a few weeks ago.
11:41:12<Sizur>c'mon oleg, don't throw us a half-baked cookie
11:41:26<mmmdonuts>E.g. I think that TABLE elements weren't implemented.
11:41:43<mmmdonuts>But take a look yourself, it wasn't hard to get it running.
11:41:56<mmmdonuts>I might have missed something.
11:42:44<mmmdonuts>Part of what I was concerned about was that even if I did extend it to support all the HTML elements I wanted, I wondered whether it would scale. The type machinery being used there scares me. :)
11:43:14<mmmdonuts>Maybe GHC could handle it, but I'm not sure if *I* could. :)
11:43:24<Sizur>yeah, it's in the Render monad
11:44:00<Sizur>so each element needs to be meaningful for all renderers?
11:45:10<mmmdonuts>Something like that.
11:45:31<mmmdonuts>I only looked at the implementation briefly.
11:46:51<Sizur>yeah, comprehending this is definitely not a piece of cake to swallow, at least for me...
11:46:55<mmmdonuts>Also, elements have to be aware of the HTML grammar they're implementing, i.e. which elements are allowed inside which others.
12:20:04<Sizur>theoretically one can use hsxml to render the same expression in html and in json, so same since code can be sent as a page or as an ajax response to populate a page
12:20:18<Sizur>fits perfectly with happs
12:29:41<mmmdonuts>yes.
12:49:38<mmmdonuts>In the google groups archives, at one point Alex mentioned that he was going to look at hsxml. I didn't see anything further about it, though.
12:50:38<mmmdonuts>I have to go now - happy templating!
12:50:43<Sizur>ttyl

Back to channel and daily index: content-negotiated html turtle