RE: tail -f behaviour in ghc.

Indeed, it appears not to work in GHC 5.00.2, but I've checked that it does work with the forthcoming 5.02. Code is attached.

Ta much -- almost exactly what I was trying ('cept for posix sleep rather than threadDelay, and other (apparent) niceties). I'll double check this against 5.02 locally, in case there's some other oddity occurring.