Stavros Macrakis wrote:
> comm.lisp works properly when compiled, but not when interpreted.
>
> This is true both for the source file that comes with 5.12.0, and for the
> current CVS file.
>
> Does anyone know why?
Neat. :-(
I took a quick look at the problem 47 in rtest16. It fails because of
an error catch. From the backtrace in cmucl, checkpow is referring to
the unknown variable num. But num is supposed to be a special var.
Perhaps declare-top is wrong?
There might be other problems. Long ago, I know James Amundson arranged
the code to be compiled and then the code to be reloaded into a fresh
lisp before dumping out a core file. I think this was to work around
some strange problem in compile order or compile-time versus load-time
or maybe even multiply-defined functions versus macros. Perhaps this is
related?
>
> Shouldn't we be testing the interpretive version of Maxima at least
> occasionally?
Good question. I always run Maxima with compiled code, even when
debugging, but mostly because cmucl's compiled code gives better results
than interpreted code.
Ray