Hi folks, I'm having trouble reasoning about laziness and FFI resources. I've written a little C wrapper function to augment Text.XML.LibXML; given a Ptr Node, it will return the node's tag-name:
foreign import ccall unsafe _getName :: Ptr Node -> IO CString Given LibXML's 'withNode' function, I can write: withNode :: Node -> (Ptr Node -> IO a) -> IO a getName1 :: Node -> IO String getName1 n = withNode n $ \n' -> _getName n' >>= peekCString But I'm tempted to write this instead: getName2 n = withNode n return >>= _getName >>= peekCString In a strict language, I'd observe that the (Ptr Node) in getName2 was escaping the lexical scope of the withNode call, and so getName2 would be unsafe (since it might get GC'd upon leaving that scope). But in Haskell, I can't reason out whether that's the case. Is 'withNode n return' a risky proposition? Is lexical scope indeed the right test here? If not, how might you reason about the safety of this code? Thanks, Graham _______________________________________________ Haskell-Cafe mailing list [email protected] http://www.haskell.org/mailman/listinfo/haskell-cafe
