[Haskell-beginners] Help me with designing my daemon, please.

[Michael Litchard]

I think I've come up with something that's more simple and makes more
sense. I will probably want to run it by the list for a sanity check.
I'll update this thread when it's ready to display.

On Wed, Sep 7, 2011 at 3:21 PM, Michael Litchard <michael at schmong.org> wrote:
> I need to provide more detail. Keep in mind that what I want to do,
> and the way I'm doing it may not fit at all. This code is using Chans
> and MVars, but could be using TChans and TVars easily enough. I kept
> things the way they are to just express what I'm trying to do
>
>> data ServerModel = PRODUCT1
>>                  | PRODUCT2
>>                  | PRODUCT3
>>                  | PRODUCT4
>>                  | PRODUCT5
>>                      deriving Show
>
>> newtype ProcessStep = PStep (ServerModel -> FilePath -> IO ())
>
>> type Cookie = String
>
>> type ProcessState = MVar ([ProcessStep],ProcessConfig)
>
>> data ProcessConfig = PConfig { model :: ServerModel
>>                              , ipAddress :: String
>>                              , rootDirectory :: FilePath
>>                              , cookie :: Cookie
>>                              } deriving Show
>
>> preProcess :: ServerModel -> FilePath -> IO ()
>> preProcess sModel fPath = putStrLn ("preProcessing" ++ show sModel)
>
>> initiatedJob :: ServerModel -> FilePath -> IO ()
>> initiatedJob sModel fPath = putStrLn ("in progress" ++ show sModel)
>
>> makeChart :: ServerModel -> FilePath -> IO ()
>> makeChart sModel fPath = putStrLn ("chart making" ++ show sModel)
>
>  main :: IO ()
>  main = do
>     pState <- make
>     world <- newEmptyMVar :: IO (ProcessState)
>     worldQueue <- newChan :: IO (Chan ProcessState)
>     installHandler userDefinedSignal2 (Catch $ worldHandler world worldQueue) N
> othing
>     installHandler userDefinedSignal1 (Catch $ emptyQueue worldQueue
> ) Nothing    ---- This is for testing purposes. I just want to be able
> to empty the Chan and see if expected behavior holds
>     installHandler nullSignal (Catch $ emptyMVar world ) Nothing
> ---- For testing as well, same reason as above.
>     sequence_ $ repeat $ queueCheck
>
>
>> worldHandler :: ProcessState -> Chan ProcessState -> IO ()
>> worldHandler world worldQueue = do
>>
>>     mvarState <- isEmptyMVar world
>>     let tStep = PStep undefined
>>     let tConfig = PConfig { model = undefined,
>>                             ipAddress = undefined,
>>                             rootDirectory = undefined,
>>                             cookie = undefined
>>                           }
>>
>>     let tState = undefined
>>     case (mvarState) of
>>       True  -> putMVar world tState
>>       False -> growQueue
>>         where growQueue = do
>>                   newWorld <- newMVar ([tStep], tConfig)
>>                   writeChan worldQueue newWorld
>
>
>
> On Wed, Sep 7, 2011 at 1:37 PM, David McBride <dmcbride at neondsl.com> wrote:
>> I'm imagining this:
>>
>> data TestInfo = {
>>  testname :: String,
>>  etc..
>> }
>> data TestResult = {
>>  success :: Bool,
>>  etc...
>> }
>>
>> data Test = Test (TestInfo -> IO ())
>> type Tests = [Test]
>>
>> main = do
>>  let tests = [whatever] :: Tests
>>  testchan <- newTChanIO :: IO (TChan TestInfo)
>>  resultchan <- newTChanIO :: IO (TChan TestResult)
>>  exceptionwhatever $ queuetest testchan
>>  forkIO $ testThread (testchan,resultchan) tests
>>  printTestResults resultChan
>>
>> queuetest chan = atomically $ writeTchan (TestInfo .....)
>>
>> testThread (testchan, resultchan) tests = forever $ do
>>  newtest <- atomically $ readTChan testchan
>>  results <- mapM tests newTest
>>  atomically $ writeTChan resultchan results
>>
>> printTestResults chan = forever $ do
>>  x <- atomically $ readTChan chan
>>  print x
>>
>> Something like that perhaps?
>>
>>
>> On Wed, Sep 7, 2011 at 3:31 PM, Michael Litchard <michael at schmong.org> wrote:
>>> This is what I am trying to do.
>>> I have tests to run and manage. I'm only running one test at a time.
>>> When my daemon gets a signal, it will either prep a test and run it,
>>> or queue the request. After it runs the test, I want it to check the
>>> queue for other tests that may have been requested.
>>> This is my first expedition into this domain. I'm trying to collect
>>> MVars and putting tem in a TChan is the way that seemed right, but I'm
>>> not sure at all. This is my first guess.
>>> I thought I needed a forked thread for the eventuality that I get a
>>> signal while my transaction is being executed.
>>> Have I clarified or further obfuscated?
>>>
>>> On Wed, Sep 7, 2011 at 12:22 PM, David McBride <dmcbride at neondsl.com> wrote:
>>>> It sounds bizarre.  Why pass around an mvar in tchan, when you could
>>>> just pass a maybe around and pattern match to see if it is Nothing or
>>>> not?  Also, why have forkio and tchan at all if they are only going to
>>>> operate in sequence, one at a time?
>>>>
>>>> What exactly are you trying to do?
>>>>
>>>> On Wed, Sep 7, 2011 at 3:01 PM, Michael Litchard <michael at schmong.org> wrote:
>>>>> I have a daemon I need to build, and need to work out some design
>>>>> details I am having difficulty with. Here's what the design looks like
>>>>> right now
>>>>>
>>>>> When the daemon starts it creates an empty MVar and an empty TChan.
>>>>> Then it listens for a usrSIG1.
>>>>> when it gets one, it checks to see if the MVar is empty. If it is, it
>>>>> does some stuff to fill the MVar, which is then used to pass around
>>>>> state for a list of functions. These functions are always the same.
>>>>> After evaluating these functions, the TChan is checked. As long as the
>>>>> TChan has something in it,
>>>>> it populates an MVar and the same three functions are evaluated in the
>>>>> same order again.
>>>>>
>>>>> If the MVar is full, it creates another MVar of the same type and puts
>>>>> it in the TChan.
>>>>>
>>>>> Is this a sound design? Does it prompt any questions from you? Here's
>>>>> my question. If this is basically a sound design, I know I will need
>>>>> use forkIO. I'm not sure where.
>>>>> If this is not a sound design, please ask questions or give other
>>>>> feedback so I can make changes and restore sanity.
>>>>>
>>>>> _______________________________________________
>>>>> Beginners mailing list
>>>>> Beginners at haskell.org
>>>>> http://www.haskell.org/mailman/listinfo/beginners
>>>>>
>>>>
>>>
>>> _______________________________________________
>>> Beginners mailing list
>>> Beginners at haskell.org
>>> http://www.haskell.org/mailman/listinfo/beginners
>>>
>>
>