> On 12/24/12 12:24 AM, Danny Yoo wrote: > > I'm currently looking at (system-type 'machine) to determine whether > > I'm on a 32-bit vs 64-bit machine. However, the strings I'm getting > > are a bit too platform specific, and I'm resorting to matching the > > string "64" somewhere in there. Is there a nicer way to approach > > this?
I don't think you want to use `(system-type 'machine)'. On Mac OS X, information in `(system-type 'machine)' tells you whether the OS kernel itself is running in 32-bit or 64-bit mode, which seems unlikely to be what you're interested in. Using `(system-library-subpath)' may be a better choice, since it depends on the Racket build instead of the kernel. The `ffi/winapi' library, for example, uses `(system-library-subpath #f)', because it's trying to detect Win64 mode specifically. I think probably we should add a 'word-size mode to `system-type', but one of the reasons that there isn't already a direct predicate is that it depends on why you want to distinguish a 32-bit from 64-bit environment. Running in 32-bit mode or 64-bit mode currently implies various things that could change in a future implementation, so maybe there should be a more direct test for what you want. Or maybe `(system-type 'word-size)' is what you need. At Mon, 24 Dec 2012 00:34:31 -0500, David Van Horn wrote: > Just a thought: (fixnum? (sub1 (expt 2 62))) That's the right test if you're interested in the range of fixnums. If you're interested in the size in bytes of a pointer for use with the FFI, then (= 8 (ctype-sizeof _pointer)) would be better, although that expression and David's currently return the same result. ____________________ Racket Users list: http://lists.racket-lang.org/users
