[Haskell-cafe] scotty + aeson + ffi: optimising performance

[Miro Karpis]

sure, sorry should have done at the first place. Basically I have several
similar calls like one below. This one should be the heaviest one.

--scotty:
post "/setmoduletable" $ do
b <- body
let queryString = tableArrayInDataToJSON b
case queryString of
Nothing -> text $ "error parsing json! "
Just (ArrayDataIn tablename columnsCnt rowsCnt values) -> do
result <- liftIO $ FM.setmoduletable tablename columnsCnt rowsCnt values
let resultInt = fromIntegral $ result::Int
json $ ReturnIntData resultInt


--json conversions
tableArrayInDataToJSON :: L.ByteString -> Maybe ArrayDataIn
tableArrayInDataToJSON rawJson = retValue
where
json = decode rawJson :: Maybe ArrayDataIn
retValue = case json of
Nothing -> Nothing
Just a -> Just a


data ArrayDataIn = ArrayDataIn{
tablename :: !String
,columnscount :: Int
,rowscount :: Int
,values :: [Double]
}deriving (Show,Generic)

instance FromJSON ArrayDataIn
instance ToJSON ArrayDataIn

--ffi

foreign import stdcall safe "setmoduletable" c_setmoduletable :: CString ->
CInt -> Ptr CDouble -> CInt -> CInt -> CInt -> IO CInt


setmoduletable :: String -> Int -> Int -> [Double] -> IO CInt
setmoduletable param columns rows array = do
let cParamLength = fromIntegral $ length param ::CInt
    cColumns = fromIntegral $ columns ::CInt
    cRows = fromIntegral $ rows ::CInt
    cTable = [realToFrac x ::CDouble | x <- array]
    firstTimeStepAfterRestart = 0::CInt
cParam <- newCString param
cTablePtr <- newArray cTable
result <- c_setmoduletable cParam cParamLength cTablePtr cColumns cRows
firstTimeStepAfterRestart
free cTablePtr
free cParam
return result




On Sat, Apr 5, 2014 at 3:01 PM, John Wiegley <johnw at fpcomplete.com> wrote:

> >>>>> Miro Karpis <miroslav.karpis at gmail.com> writes:
>
> > I tried to profile it but so far I can see that the bottleneck is Aeson
> > (please correct me if I'm wrong). Is there a way how to optimise it? The
> > data I'm sending via JSON are very small, so that can not be the problem.
>
> Can you share any of your code with us?
>
> John
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe at haskell.org
> http://www.haskell.org/mailman/listinfo/haskell-cafe
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/haskell-cafe/attachments/20140405/3566ec82/attachment.html>