"S. Alexander Jacobson" <[EMAIL PROTECTED]> wrote:

> Given that we have no easy way to serialize thunks, the whole RPC 
> approach just seems wrong for Haskell.

Tom Shackell is developing a simple bytecode reflection API to be
implemented in the yhc compiler.  This will allow the transmission of
arbitrary function code across the network for execution on a remote
machine, with the results fed back to the calling machine.

e.g.
    encodeF :: (a->b) -> Bytecode (a->b)
    dynload :: Bytecode (a->b) -> (a->b)

Work in progress...  There are doubtless lots of security issues here,
not least type safety....

Regards,
    Malcolm
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe

Reply via email to