From: William H. N. <wil...@ai...> - 2001-03-07 14:48:16
|
On Wed, Mar 07, 2001 at 02:19:00PM +0100, Martin Atzmueller wrote: > I ran into a problem using the bleeding edge CVS version of sbcl. > Well, bare sbcl-0.6.11 works, so this bug must have been introduced > later. > > Problem: > If I compile and load mk-defsystem (attached), I get > > debugger invoked on condition of type TYPE-ERROR: > TYPE-ERROR in SB-KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: > 0 is not of type (OR FUNCTION SB-KERNEL:FDEFN). I can replicate this problem with 0.6.11.12, but I haven't tried to figure it out yet. I've wandered into a mess with what should become 0.6.11.13, where I've tried to enable a lot more of the new INTERSECTION-TYPE stuff, and I'm having trouble getting it to boot; and I've done too much work on it to want to back out and start over. I'm not sure how long it will take to get that into a stable state (or for me to give up and start over at 0.6.11.12). After I do, I'll try to find out what's wrong with the byte interpreter. > Since you said on the mailing list, that you would continue to work on > that, this bug may have been fixed by now. > Otherwise, I hope you can make some sense out of this. Unless the byte interpreter problem is a weird side effect of some type system bug, I doubt that the changes in 0.6.11.13 will fix it. But I can try to fix it after that. In the meantime, a possible workaround is (compile-file "defsystem-sbcl-1.lisp" :byte-compile nil) (load *) When I tried that, I got a more familiar error: error in function SB-C::%DEFCONSTANT: The constant *FIND-IRIX-VERSION-SCRIPT* is being redefined. -- William Harold Newman <wil...@ai...> software consultant PGP key fingerprint 85 CE 1C BA 79 8D 51 8C B9 25 FB EE E0 C3 E5 7C |
From: Martin A. <ma...@at...> - 2001-03-07 17:54:22
Attachments:
bytecode-patch-20010307.diff.gz
|
William Harold Newman wrote: > > On Wed, Mar 07, 2001 at 02:19:00PM +0100, Martin Atzmueller wrote: > > I ran into a problem using the bleeding edge CVS version of sbcl. > > Well, bare sbcl-0.6.11 works, so this bug must have been introduced > > later. > > > > Problem: > > If I compile and load mk-defsystem (attached), I get > > > > debugger invoked on condition of type TYPE-ERROR: > > TYPE-ERROR in SB-KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: > > 0 is not of type (OR FUNCTION SB-KERNEL:FDEFN). > > I can replicate this problem with 0.6.11.12, but I haven't tried to > figure it out yet. I think I've found out what was wrong. The bug was introduced by the generate-byte-code-set patch, and the "delete the duplicate dup" tweak. A patch is attached. -- Martin Atzmueller <ma...@at...> |
From: William H. N. <wil...@ai...> - 2001-03-14 13:47:10
|
On Wed, Mar 07, 2001 at 07:55:01PM +0100, Martin Atzmueller wrote: > William Harold Newman wrote: > > > > On Wed, Mar 07, 2001 at 02:19:00PM +0100, Martin Atzmueller wrote: > > > I ran into a problem using the bleeding edge CVS version of sbcl. > > > Well, bare sbcl-0.6.11 works, so this bug must have been introduced > > > later. > > > > > > Problem: > > > If I compile and load mk-defsystem (attached), I get > > > > > > debugger invoked on condition of type TYPE-ERROR: > > > TYPE-ERROR in SB-KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: > > > 0 is not of type (OR FUNCTION SB-KERNEL:FDEFN). > > > > I can replicate this problem with 0.6.11.12, but I haven't tried to > > figure it out yet. > > I think I've found out what was wrong. > The bug was introduced by the generate-byte-code-set patch, and the > "delete the duplicate dup" tweak. > A patch is attached. The patch has been merged as part of 0.6.11.14, just checked in. -- William Harold Newman <wil...@ai...> software consultant PGP key fingerprint 85 CE 1C BA 79 8D 51 8C B9 25 FB EE E0 C3 E5 7C |