Re: Death by pointers. (Was: order of argument

> Actually, of course, they were wrong, but nevertheless I
>think we can safely say that nobody knows what the programming
>langauge of 2010 will look like, but it will be called something
>beginning with C.
>
> Adrian
>[Cobol. -John]
(Ouch!)


Actually, I suspect that "the language of 2010" is going to be very
different. [Oh, no, he's about to start talking about his thesis
topic...] We will probably have the technology to be building quantum
computers by then (computers which can be in superpositions of
squillions of different states at once.)


Optimisation will be a very weird business indeed. You can't observe
the contents of a variable until the computation is finished, because
that observation will affect the computation. On the other hand, you
can always "undo" the last operation that you did, right back to the
beginning of the program.


A really clever compiler will need to work out what things can be
placed in "classical" memory, and what things really need to be
quantum parallelised. And then the loops in quantum code need to take
exactly the same number of steps regardless of input, otherwise they
will break syncronisation. So an optimising compiler might realise
that the best way to render one particular section of code is to
quantum parallelise it, which might then involve pessimising certain
sections so that they take longer!
______________________________________________________________________


            Gregory D. Baker (gbaker@rp.csiro.au) CSIRO Radiophysics
______________________________________________________________________


        PROBABLE (http://www.rp.csiro.au/~gbaker/probable/index.html)
                - Predictions of the future ... read some ; add some!
          Q-gol (http://www.rp.csiro.au/~gbaker/q-gol/index.html)
                - High-level access language for quantum computers
--

Return to the comp.compilers page.
Search the comp.compilers archives again.