Hello,
Using sbcl 0.9.11 (+ threads) with slime from cvs duplicates output rather frequently.
Interestingly, this is a problem only on my single cpu machine (pentium III Mobile) and there it hapens very often. On the one with two cpus (amd64 x2), if it happens with slime from cvs and sbcl 0.9.11 with threads, it must be very rare; I haven't seen it yet.
Any ideas?
Regards, Mario.
The problem, it seems, is that the fix that was commited the 12th of april only works if *use-dedicated-output-stream* is nil.
At the moment, my plan for the distribution tar balls is to just add the site-init.lisp file with
#+sbcl (setf swank::*use-dedicated-output-stream* nil)
While this works and solves the problem, it probably isn't the best posibility.
Does anyone have a better idea?
Regards, and thanks,
Mario
Mario S.Mommer m_mommer@yahoo.com writes:
Hello,
Using sbcl 0.9.11 (+ threads) with slime from cvs duplicates output rather frequently.
Interestingly, this is a problem only on my single cpu machine (pentium III Mobile) and there it hapens very often. On the one with two cpus (amd64 x2), if it happens with slime from cvs and sbcl 0.9.11 with threads, it must be very rare; I haven't seen it yet.
Any ideas?
Regards, Mario.
* Mario S.Mommer [2006-04-19 22:16+0200] writes:
The problem, it seems, is that the fix that was commited the 12th of april only works if *use-dedicated-output-stream* is nil.
At the moment, my plan for the distribution tar balls is to just add the site-init.lisp file with
Please don't distribute files which aren't in CVS.
#+sbcl (setf swank::*use-dedicated-output-stream* nil)
While this works and solves the problem, it probably isn't the best posibility.
I switched the dedicated stream off by default. There are also race conditions on the Emacs side and it's easier to connect to remote machines if only one open port is needed.
Helmut.