On Wed, Dec 07, 2005 at 05:05:58PM -0000, Simon Marlow wrote:
No, that is exactly what I disagree with. Faced with non-derministic semantics, the compiler does *not* have to preserve all possible outcomes. In other words, it does not have to assume that an IORef can be modified between any two arbitrary instructions. If we had to assume this, then indeed all bets would be off, and C compilers would be very much more restricted in what optimisations they could perform.
Yup. this is exactly why C has the 'volatile' keyword. variables that are declared volatile will always be read and written to memory and never stored in a register because they could be modified by external interrupts or threads. If every varibale were considered volatile everything would be a whole whole lot slower. I have only had need to use it 3 times and all were in an operating system kernel. In any case, ghc IORefs are not 'volatile' in the C sense and that is a good thing. John -- John Meacham - ⑆repetae.net⑆john⑈