Instead of using the FFI, you can also write a library that wraps the
"curl" command-line executable. This has the advantages of keeping
large file I/O out of Racket (lower processing overhead, no GC,
schedulable by the host OS on other CPUs/cores), and of process
isolation (i.e., not having additional C code in the same process as the
Racket VM, so that less risk of the C code causing stability or security
programs).
(There is a somewhat fancy "curl" process Racket library that someone
wrote, but I don't know whether or when they'll be able to open source
it. Probably best to just write a simple one that does what you need.)
Neil V.
--
http://www.neilvandyke.org/
____________________
Racket Users list:
http://lists.racket-lang.org/users
