MS> Another problem with this approach is that it breaks in the MS> following case: MS> (setq bar 23) MS> (defmacro foo () bar) MS> (defun baz () (foo)) MS> (baz) => 23 MS (setq bar 42) MS> (baz) => [still ...] 23 MS> This makes obvious that the macro dereferences the binding too MS> early. (Namely, at expansion time.) Time seems to getting near to issue a bug-report to Bruno Haible[1] ... (again the following is from ACL4.3): USER(8): (setq bar 23) 23 USER(9): (defmacro foo () bar) FOO USER(10): (defun baz () (foo)) BAZ USER(11): (baz) 23 USER(12): (setq bar 42) 42 USER(13): (baz) 42 No surprise here: (foo) => 42 ! USER(20): (setq bar 23) 23 USER(21): (let ((bar 42)) (baz)) 23 USER(22): (setq bar 57) 57 USER(23): (let ((bar 42)) (baz)) 57 USER(24): (foo) 57 Footnotes: [1] .. but of course, we would have to check the HyperSpec at least to see if ACLs behaviour is correct (required by the ANSI definition that is). Holger PS: I begin to see the problems that you have with CL-macros. That won't make me switch to Scheme, however. I have lived up to avoid these problems. -- --- http://www.coling.uni-freiburg.de/~schauer --- ">grep schauer passwd "Das ist nur eine Shell. Bedenklich waere, wenn Dein LILO /vmunix.el bootet." -- Kristian und Holger in de.comp.os.linux.misc