Re: I must be going mad!

"Stephen J. Turnbull" <stephen(a)xemacs.org&gt; writes: PLACE wasn't specified as a literal, it just happened to refer to a literal. There is no way a macro can generally determine whether (push x (cdr y)) modifies a cons received through a literal or not. If you really wanted to "fix" this, you could: a) proclaim modification of list literals /kosher/. This would require removing the optimization that reuses literals by storing them to the compiled-function object. or, you could b) make '(...) create read-only conses and have setcar/setcdr check for such conses. In theory, this slows down setcar and setcdr, but I'm not sure if it would be a slow-down for us, since we have code that checks for read-only conses anyway. But it would require special handling of QUOTE in a way I don't like. (Currently QUOTE simply returns what the reader gave it.) Both "solutions" require slowdowns or complications similar to what would be needed to allow my sample C construct to work as expected. I don't think it's a problem worth solving.