I think this relates to a known error consisting unexpectedly closed connections:

http://common-lisp.net/pipermail/slime-devel/2009-December/016985.html

I've experienced this too [1], but in a strange context.

Slime works fine for me on OS X 10.4 and 10.5 on a G4 powerbook when running in

1) Aqua emacs 2) OS X terminal emacs 22.3 3) normal X11 emacs 22.3 4) emacs 22.3 in an X11 xterm

But I like to run Xephyr, an X server that runs inside another X11 in its own big window. For me, it interacts with Aqua spaces better than Apple X11, allowing a virtual window manager within a single Aqua space. So I have X11 running, and Xephyr running inside it. Works fine, mostly. But now the strange things happen:

5) slime works in an xterm emacs in Xephyr 6) slime often does not work in an X11 emacs 22.3 running in Xephyr, giving error [1] below

This failure occurs with greatest likelihood when a large compilation is occurring, like a fink (OS X unix software) update. When the update stops, slime works again. The failure occurs equally for SBCL and CCL. The failure does not occur in an emacs running in an xterm, even inside Xephyr. I've tried increasing the system limits, including processes, memory, and open files, but it doesn't seem to help.

Hence it looks a bit like it might be a subtle timing issue connected with running an inherently slow X11 emacs slowed further by the Xephyr layer while the system is bogged down doing other things.

I hope that this email at least helps to outline the boundaries of this problem.

-JK

======= [1] - close message: ;; Swank started at port: 49194. 49194 * +++ SWANK-PROTOCOL-ERROR: nil #<swank::swank-protocol-error {127BA821}> ;; swank:close-connection: end of file on #<sb-sys:fd-stream for "a socket" {125 AA421}> =======