[Haskell-iPhone] Automatically freeing function pointers

[David Pollak]

On Mon, Jun 20, 2011 at 1:46 PM, Stephen Blackheath [to GHC-iPhone] <
likeliest.complexions.stephen at blacksapphire.com> wrote:

> David,
>
> I hadn't quite grasped your code. The problem with your code is that these
> two lines...
>
> >                                     func <- funky
> >                                     freeHaskellFunPtr func
>
> ...allocate then deallocate a function pointer, with a net result of
> nothing useful, and no effect on the function pointer you allocated on the
> third line.
>
> If you use this trick, however...
>
> {-# LANGUAGE DoRec #-}
>
> foreign import ccall safe "wrapper" mkInit
>    :: (Ptr WorkerThread -> CInt -> IO ()) -> IO (FunPtr (Ptr WorkerThread
> -> CInt -> IO ()))
>
> applicationMain init dealloc = mdo
>    fInit <- mkInit $ \callbackWorker fd -> do
>        freeHaskellFunPtr fInit
>        ...
>    ...
>
> ...then you can do it quite painlessly.  Note that DoRec is correct for ghc
> 6.10, but its name has changed in later ghc versions.  The compiler will
> tell you.
>


Excellent!  Thanks for the "pointer" oh wait... "thunk" oh wait... ;-)


>
>
> Steve
>
>
> On 21/06/11 05:00, David Pollak wrote:
>
>>
>>
>> On Sun, Jun 19, 2011 at 1:32 PM, Stephen Blackheath [to GHC-iPhone]
>> <likeliest.complexions.**stephen at blacksapphire.com<likeliest.complexions.stephen at blacksapphire.com>
>> <mailto:likeliest.complexions.**stephen at blacksapphire.com<likeliest.complexions.stephen at blacksapphire.com>>>
>> wrote:
>>
>>    David,
>>
>>    In GHC-iPhone the number of function pointers that are allowed to be
>>    active for a given function is limited, but you can change the size
>>    of the limit.  This is documented in
>>
>>
>> Steve,
>>
>> Yep... I understand that there are not an unlimited number of function
>> pointers.  However, the function that I wrote was "supposed" to
>> automatically release the function:
>>
>> runOnMain :: IO () -> IO ()
>> runOnMain todo = do
>>                   func <- funky
>>                   dispatchFunc func
>>   where funky =  mkStrCB $ \v -> do
>>                                    todo
>>                                    func <- funky
>>                                    freeHaskellFunPtr func
>>
>> Note the freeHaskellFunPtr function call at the end of the lambda.
>>
>> I'm thinking that the "func <- funky" line is being treated as a
>> recursive call and I'm also thinking that this is a limitation of my
>> understanding the magic of lazy evaluation.
>>
>> So, please let me rephrase my question:
>>
>> How do I get the pointer to a wrapper function so that the pointer can
>> be used within the wrapped function itself?
>>
>> I have worked around the particular issue by making a call to
>> freeHaskellFunPtr in the Objective-C code that invokes the function.
>> It's less optimal that being able to have the Haskell code be able to
>> auto-free the function pointer after the function is invoked.
>>
>> Thanks,
>>
>> David
>>
>>    http://projects.haskell.org/__**ghc-iphone/downloads/GHC-__**
>> iPhone.pdf<http://projects.haskell.org/__ghc-iphone/downloads/GHC-__iPhone.pdf>
>>    <http://projects.haskell.org/**ghc-iphone/downloads/GHC-**iPhone.pdf<http://projects.haskell.org/ghc-iphone/downloads/GHC-iPhone.pdf>
>> >
>>
>>    on p5 under 'Pool sizes for foreign function "wrapper" callbacks'.
>>
>>
>>    Steve
>>
>>
>>    On 18/06/11 11:29, David Pollak wrote:
>>
>>        Howdy,
>>
>>        I tried to build a function that would create an auto-freeing
>>        function
>>        pointer:
>>
>>        runOnMain :: IO () -> IO ()
>>        runOnMain todo = do
>>                           func <- funky
>>                           dispatchFunc func
>>           where funky =  mkStrCB $ \v -> do
>>                                            todo
>>                                            func <- funky
>>                                            freeHaskellFunPtr func
>>
>>        Unfortunately, I wind up getting:
>>          internal error: iPhoneCreateAdjustor - adjustor pool
>>        'Main_d2K2' is
>>        empty (capacity 32)
>>             (GHC version 6.10.4-iphone-simulator for i386_apple_darwin)
>>
>>        I'm not sure why, but I suspect that funky is being called
>>        recursively
>>        and thus stuff is never being freed.
>>
>>        Any hints as to the issue and solutions?
>>
>>        Thanks,
>>
>>        David
>>
>>
>>        --
>>        Lift, the simply functional web framework http://liftweb.net
>>        Simply Lift http://simply.liftweb.net
>>        Follow me: http://twitter.com/dpp
>>        Blog: http://goodstuff.im
>>
>>
>>
>>        ______________________________**___________________
>>        iPhone mailing list
>>        iPhone at haskell.org <mailto:iPhone at haskell.org>
>>
>>        http://www.haskell.org/__**mailman/listinfo/iphone<http://www.haskell.org/__mailman/listinfo/iphone>
>>        <http://www.haskell.org/**mailman/listinfo/iphone<http://www.haskell.org/mailman/listinfo/iphone>
>> >
>>
>>
>>    ______________________________**___________________
>>    iPhone mailing list
>>    iPhone at haskell.org <mailto:iPhone at haskell.org>
>>
>>    http://www.haskell.org/__**mailman/listinfo/iphone<http://www.haskell.org/__mailman/listinfo/iphone>
>>    <http://www.haskell.org/**mailman/listinfo/iphone<http://www.haskell.org/mailman/listinfo/iphone>
>> >
>>
>>
>>
>>
>> --
>> Lift, the simply functional web framework http://liftweb.net
>> Simply Lift http://simply.liftweb.net
>> Follow me: http://twitter.com/dpp
>> Blog: http://goodstuff.im
>>
>>


-- 
Lift, the simply functional web framework http://liftweb.net
Simply Lift http://simply.liftweb.net
Follow me: http://twitter.com/dpp
Blog: http://goodstuff.im
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/iphone/attachments/20110620/8ab805ae/attachment-0001.htm>