Haven't looked at the code, but off the top of my head the ones mentioning offset problems might be from too-large classes. Does the system compile very large functions?

The last trace looks like it might be related to character encoding mismatch - perhaps invalid utf-8?

Alan

On Mon, Dec 4, 2017 at 2:33 PM, Kurt Pagani <nilqed@gmail.com> wrote:

I (almost) successfully compiled the fricas computer algebra system using ABCL (https://github.com/nilqed/fricas0).
"Almost" because four of the over 1300 algebra files are refractory. I essentially got two errors:

   >> System error:
   The value :INT is not of type (COMMON-LISP:MEMBER COMMON-LISP:NIL
                                                  :BOOLEAN).

and

   >> System error:
   The assertion (COMMON-LISP:<= -32768 JVM::OFFSET 32767) failed.

While the first one might be related to

http://abcl.org/trac/ticket/316 (http://abcl.org/trac/changeset/15034)
http://armedbear-devel.common-lisp.narkive.com/rm9OArfe/the-value-long-is-not-of-type-member-nil-boolean

I have no idea what the second one could mean. Google, however, shows similar ones @:
https://common-lisp.net/project/cl-test-grid/ql/qlalpha-2014-08-23-abcl-load-failures.html (fishpack, lapack ...).

Well, I'm stuck, so any ideas would be very welcome.
Many thanks
Kurt