- web-devel - Haskell.org

WAI, HTTP POST , POST request body extract
by Антон Чешков 06 Mar '11

06 Mar '11

Hi Michael. Please explain how to extract POST data from WAI Request. Thanks. -- Best regards, Cheshkov Anton Phone: +7 909 005 18 82 Skype: cheshkov_anton

1 1

Having trouble sorting through the code for haskellers.com
by Michael Litchard 03 Mar '11

03 Mar '11

I tried running the codebase, and got this error. runhaskell fastcgi.hs fastcgi.hs: SqlError {seState = "", seNativeError = 1, seErrorMsg = "connectPostgreSQL: FATAL: password authentication failed for user \"haskellers\"\nFATAL: password authentication failed for user \"haskellers\"\n"} That's fine. I just need to know what the expected password should be, right? Then I can set up the postgres user Haskellers properly. Except I can't find exactly where a password/username are being passed. Telling me which file I should stare at until enlightenment comes would be helpful.

2 2

WAI, http-enumerator, json-enumerator
by Антон Чешков 03 Mar '11

03 Mar '11

Hello All. Could anyone provide simple wai application example to better understand json-enumerator and http-enumerator works. This application must make: 1. takes POST request with json data "{a : 0}" 2. extract json data and with json-enumerator change value of "a" from 0 to 1 3. send post request to URL (for example http://bla-bla.com) with http-enumerator with json data "{a : 1}" 4. recieve response from http://bla-bla.com 5. send response with json data "{a : 1}" Thanks. -- Best regards, Cheshkov Anton Phone: +7 909 005 18 82 Skype: cheshkov_anton

1 0

Fwd: [Haskell-cafe] type safe web forms at compile time Was: Re: Ur vs Haskell
by Alberto G. Corona 03 Mar '11

03 Mar '11

forward to web-devel: WASH is full of wonderful ideas . packed in a not so wonderful syntax. It is worth to evolve it. WASH does force form safety in a similar way to Formlets <http://hackage.haskell.org/package/formlets>: because the form and the form read code are generated automatically by a class instance. So there is no need for, type checking safety The problem comes when the form is generated and/or maintained( edited by some people (some forms have lot of formating in the real world) while the form handling code is .maintained by some other people, the programmers. In this real case , type cheching is very important. 2011/3/2 Marc Weber <marco-oweber(a)gmx.de> Excerpts from Alberto G. Corona's message of Wed Mar 02 20:53:28 +0000 2011: > > Some time ago I forgot to forward this message to thie ur versus Haskell > > <http://www.haskell.org/pipermail/haskell-cafe/2011-January/088060.html > >discussion, > > (as usual) > > --- > > The most impressive feature (of ur) is the compile time checking of > > conformance between the form and the form results. This can be done in > > See WASH (-> hackage). So there is a Haskell implementation. There are > small > differences though: urweb has nicer URLS which should be much more SEO > friendly. > > From my point of view its not only about forms. Its also about checking > SQL queries. And urweb seems to do this very well. > > Marc Weber >

2 1

type safe web forms at compile time Was: Re: [Haskell-cafe] Ur vs Haskell
by Alberto G. Corona 02 Mar '11

02 Mar '11

Hi. Some time ago I forgot to forward this message to thie ur versus Haskell <http://www.haskell.org/pipermail/haskell-cafe/2011-January/088060.html>discussion, (as usual) --- The most impressive feature (of ur) is the compile time checking of conformance between the form and the form results. This can be done in Haskell using HList magic and some class instances, I guess. ---- Since then I have been playing mentally with this. Recently I found something simple an interesting enough to share. (Although crude). It is a kind of typed form fields data Input a= Input String Type a (String -> Either String a) and a kind of heterogeneous list to aggregate form fields and results with the operator (:*): Input a :* Input b ;* Input c.... a :* b :* c and a (simulated for the purpose of demonstration) send-receive function that type match the form fields and the results: *Main> let form = Input "" Text "stringdata" novalidate :* Input "" Text (1::Integer) novalidate *Main> ask form >>= \(a :* b) -> return $ a ++ b <interactive>:1:0: No instance for (FormDigest (Input [Char] :* Input Integer) ([a] :* [a])) ...... notifying that there is no translation defined , because the result requires two lists of the same type when the form gives a string and an Integer But forcing the correct monomorphic types it does pattern match and return the values. *Main> ask form >>= \ (a :* b) -> print ('s':a) >> print ( fromInteger $ b) "sstringdata" 1 ask is just a simulation of HTTP one time interaction. It returns the input values. The whole loop involves the rendering of the form, with render: *Main> render form <input type="Text" name="var1" value="stringdata"/> <input type="Text" name="var2" value=1/> In a real case the results are read and validated from the the post values.They are (or can be) ordered sequentially acording with Input field names. The FormDigest instances do this work. There is no need to define new FormDigest instances. (although non one to one field-result can be created) The text is in literate haskell. There is a more elaborate example at the end. I know that the instances are non tail recursive and there are factorization pending but this is just a proof of concept: > {-# LANGUAGE > FlexibleInstances, > MultiParamTypeClasses, > TypeOperators > #-} > import Control.Monad.State The Heterogeneous list agregator. I tried to use GADTs but they does not easily pattern match > data x :* xs = x :* xs deriving (Read, Show, Eq) > infixr 5 :* > data Type= Hidden | Text deriving (Show, Read) -- to add all the types the input field, with text formatting, initial value and runtime validator > data Input a = Input String Type a (String -> Either [String] a) > instance(Show a)=> Show (Input a) where > show (Input _ _ x _) = show x rendering of the form need a sequentiation of field names. I use a state monad for this > class RenderForm a where > renderForm :: a -> State Int String > instance (Show a) => RenderForm (Input a) where > renderForm input = do > s1 <- render1 input > n <- get > put $ n + 1 > return s1 HList school here: > instance (Show a,RenderForm xs) => RenderForm (Input a :* xs) where > renderForm (input :* xs)= do > n <- get > put $ n+1 > h <- render1 input > s <- renderForm xs > return $ h++s > render1 (Input msg t x _)= do > n <- get > put $ n+1 > return $ msg > ++ "<input type=\"" > ++ show t ++ "\" name=" > ++ "\"var"++show n ++ "\" value=" > ++ show x ++"/>\n" > > render form= putStrLn $ evalState (renderForm form ) 0 processing of the returned form result, in an ordered String list, according with seuquential names of the fields defined in renderForm. > class FormDigest a b where > formDigest :: a -> [String] -> Either [String] b "Input a" is diggested into a type "a" > instance FormDigest (Input a) (a) where > formDigest (Input _ _ x f) (s: ss)= case f s of > Right x -> Right $ x > Left x -> Left x recursively add instances for any list of inputs Input a's are diggested into a's > instance FormDigest as bs > => FormDigest (Input a :* as) (a :* bs) where > formDigest (input :* fs) es@(s:ss) = > let er = formDigest fs ss > e = formDigest input es > in case (e, er) of > (Right x, Right ys) -> Right $ x :* ys > (Right _, Left errs) -> Left errs > (Left err, Left errs) -> Left (err ++ errs) simulated request-response that returns the entered input values > sendRec xs= do > let strs = showValues xs > return $ formDigest xs strs > class ShowValues a where > showValues :: a -> [String] > instance Show x => ShowValues (Input x) where > showValues i = [show i ] > instance (Show x,ShowValues xs) => ShowValues (Input x :* xs) where > showValues (i :* xs)= show i : showValues xs end of simulated request response > ask :: (ShowValues a, FormDigest a b) => a -> IO b > ask form = do > er <- sendRec form > case er of > Left errs -> error "" -- shoud be: "ask1 errs form "·to render form and errors > Right res -> return res EXAMPLE: > data Emp= Emp{name::String, salary :: Int} deriving Show > emp= Emp "Luis" 10000 toy html operators: > b msg = ("<b> " ++ msg ++ " </b>\n") > p msg = ("<p> " ++ msg ++ " </p>\n") > novalidate n= Right $ read n > main= do > > let form = Input > ( b "please enter the name" > ++ p "mas texto") > Text (name emp) novalidate > :* Input > (b "please enter the salary" > ++ p "jkjkjk") > Text (salary emp) novalidate > render form the matching thing > (n :* s ) <- ask form > print emp > print $ Emp n s

1 0

Re: [web-devel] [Yesod] Teach me how to write to access
by Greg Weber 01 Mar '11

01 Mar '11

The default scaffolding site creates a connection pool at startup and stores it in the foundation type. You can access the foundation type in your handler with: y <- getYesod Look at the runDB function. It accesses the connection pool from the foundation type. Without knowing any more specifics of your situation, you probably just want to write a function runDB2 that runs against a different connection pool (to your other database) that you also create at startup. Greg Weber Hi > This might not be the correct answer to your problem, but I asked about how to pass a hashtable to the application in a prevoius mail. > The answer was to pass the ioref in the data constructor - Michael wrote a very good example at: https://github.com/snoyberg/haskell-web-rosetta/blob/master/counter/yesod.hs > It might be you can connect to the database, and then pass it on and handle this second db-connection manually? > Cheers, > Petter Egesund > On Mon, Feb 28, 2011 at 10:05 AM, Katsutoshi Itoh <cutsea110(a)gmail.com> > wrote: > Hi > > I write an application, which must to access 2 databases. > 1st is main database for this application. > 2nd database is used as main database by another application. > > Could you teach me how to write to access 2nd database? > Since there is no access to the 2nd database only partially, I'm a > little problem in an ad hoc way.

1 0

[Yesod] Teach me how to write to access second database.
by Katsutoshi Itoh 28 Feb '11

28 Feb '11

Hi I write an application, which must to access 2 databases. 1st is main database for this application. 2nd database is used as main database by another application. Could you teach me how to write to access 2nd database? Since there is no access to the 2nd database only partially, I'm a little problem in an ad hoc way.

2 1

ANN: Hakyll 3
by Jasper Van der Jeugt 28 Feb '11

28 Feb '11

Hello all, I've just uploaded the 3.0.0.0 version of Hakyll [1] to Hackage [2]. This is a complete rewrite, and completely backward-incompatible with previous versions. Sorry for that. On the other hand, I believe almost all aspects of the library have improved tremendously. Hakyll now uses a declarative EDSL in Haskell to specify compilation rules, and an arrow-based EDSL to describe the compilation process. If you want to have a quick look, I recommend have a glance at the example in the tutorial [3]. More documentation needs to be written, but I wanted to make sure I can incorporate your feedback into that process. [1]: http://jaspervdj.be/hakyll/ [2]: http://hackage.haskell.org/package/hakyll-3.0.0.2 [3]: http://jaspervdj.be/hakyll/tutorial.html Cheers, Jasper

1 0

[WAI] Switching to WAI/Warp...
by Bardur Arantsson 27 Feb '11

27 Feb '11

Hi all, I've recently switched my "Hums" UPnP server over to using WAI/Warp, but I'm seeing quite high CPU usage compared to monadic I/O. (1-7% with WAI as opposed to 0-1% for monadic). (Just to be clear: The server is using a constant amount of memory, so it's definitely not leaking or anything like that.) I think the source of the efficiency problem is that I'm enumerating strict ByteStrings from a file, but the WAI response enumerator requires Builder chunks, so I'm forced to use "fromByteString" to convert between the strict ByteString representation and the Builder representation -- I'm assuming this performs a memory copy. Is there any way to avoid this extra "fromByteString"? Other than this little issue, WAI seems to be working very well and it seems like a great fit if you just want low-level access to the HTTP protocol. On a slightly related note: Is there an elegant way to map enumerator chunk elements from type a -> b using a mapping function? Right now, I've embedded the ByteString -> Builder conversion deep in my enumerator, but it should really be happening at a higher level. Cheers, Bardur

3 8

Problem with multipart request on yesod/wai
by Nubis 27 Feb '11

27 Feb '11

Hi guys, I'm developing a webservice in Yesod that needs to accept multipart POST's with image data in them. This webservice will be used mainly by mobile devices. My handler that takes care of receiving the image works correctly when receiveng a POST crafted by google chrome (from a multipart/form-data encoded form), but when receiving a multipart post from a mobile device it fails to parse the multipart request arguments. I've setup a ruby on rails application with a similar handler, and both request from my mobile device and from a browser work, so I'm leaning towards thinking this is a yesod/wai issue. This is an excerpt from my handler, that I use to debug this: myHandler = do (params, files) <- runRequestBody liftIO $ putStrLn $ "params " ++ show params When requesting from the mobile device it just shows an empty list. I have dumps of both multipart requests, they seem pretty similar with the main differences being: * the multipart boundaries are longer in the browser request. * the request from the mobile device adds a trailing semicolon to the content-type of each part (except for the image) I attached the request that's not parsing correctly Any help will be greatly appreciated, thanks in advance! --nubis :)

4 5