Re: Bug in error-reporting code
Gary Byers <[email protected]> Fri, 28 Oct 2005 05:53:38 -0600 (MDT)
| Newsgroups | gmane.lisp.openmcl.bugs |
|---|---|
| Message-ID | <[email protected]> |
At the point where vm/objects is loaded: ? (find-symbol "METHOD" "VM") METHOD :INHERITED ? (symbol-package *) #<PACKAGE "COMMON-LISP"> vm/objects.lisp contains: (defclass method ...) which has the effect of redefining the system class CL:METHOD. Not too surprisingly, things go downhill (rapidly) from there; in particular, the error system has difficulty invoking any (CLOS) methods, and notices that something is very, very, wrong ... You probably want to shadow the symbol METHOD in your VM package definition, so that CL:METHOD and VM::METHOD are distinct. The implementation probably should try to make it harder to (accidentally) redefine critical built-in classes. On Fri, 28 Oct 2005, [ISO-8859-1] Sebastián González wrote: > Hi, > > The attached code uncovers a bug in OpenMCL, both in the 32 and the 64-bit > Darwin PPC 1.0 versions: > > $ openmcl -l main.lisp > > Bug in OpenMCL system code: > Error reporting error > make: *** [run-openmcl] Error 137 > > The error seems not to be critical. > I hope the attached code proves useful in spotting and fixing the bug... > > Thanks! > Sebastián > >  > P.S. > >> The lists here get anywhere from a few dozen to a few hundred spam >> messages a day > > May be messages containing the words "bug, error, problem" shouldn't be wiped > out without being checked first... > > ------------------------------------------------------- > Sebastián González > E-mail: [email protected] > Web: http://www.info.ucl.ac.be/~sgm/ > Tel: +32 10 47 91 08 > Département d'Ingénierie Informatique > Université catholique de Louvain (UCL) > Belgium > ------------------------------------------------------- > > > > > _______________________________________________ Bug-openmcl mailing list [email protected] http://clozure.com/mailman/listinfo/bug-openmcl