Calling these tools can be a little more complicated.
I'd put those port close calls in a "dynamic-wind" cleanup thunk. I'd also catch the exceptions that could be raised by the port-closing procedure itself in the cleanup thunk. And I'd make sure that any custodian I was using didn't close the ports at the wrong time for me.
Regarding I/O, I'd make stdin a null port so that the process wouldn't block (unless I actually had to feed it input). And I'd probably poll the ports with "sync", so that, for example, I get any potential error messages to stderr, and so that that the process's stderr buffer filling won't block the process while I'm just reading the process's stdout.
I also prefer to use "process*" over "process", because it eliminates some possible command line assembling errors (especially quoting and escaping) and eliminates resulting exploits. (I'll just assert here that most people who assemble command lines as single strings using variables do it wrong and potentially create security exploits or accidents.)
Neil V. -- http://www.neilvandyke.org/ ____________________ Racket Users list: http://lists.racket-lang.org/users
