hi!

did anybody notice that when using slime with sbcl starting from a custom core file, then the communication channel between the repl buffer and sbcl seems to be buffered? (sometimes i need to press an enter for the ucw log output or compile output to arrive in emacs) i don't seem to experience this if i start a vanilla sbcl and (require ...) the fasl's into it, but it takes ages compared to starting from a core file.

i create my sbcl core file from xemacs, so swank is already loaded when the core file is saved, and my guess is that when the communication channel is re-estabilished between slime and swank, then the flush mode on the unix file handle is not restored properly. (i have no experience with how sbcl handles external resources)

unfortunately i can't test right now what happens when not loading swank into the core file, but i'll try it tonight.

- attila

(alias 101 on irc &no 'its not lisp code :)