David Steuber writes: > I'm wondering just how afraid I should be of "interesting > behaviour". It sounds like something shockingly bad ;-) Unless you're calling/writing FFI stuff that makes system calls and doesn't handle EINTR properly, I expect you shouldn't run into anything really bad : as the doc says, the streams code is believed to handle this properly already. Don't feel you're being starved of excitement, though: there are other and more exciting thread-related bogosities deep in the SBCL system that you might run into. See Robert Marlow's post, for example > Anyway, I'm going to try and hit this stuff real soon now with a Lisp > app I am trying to work on. Linux x86 with 2.6.5 kernel. Keep us informed... -dan -- "please make sure that the person is your friend before you confirm"