Thanks for your reply.
I am well aware of the meaning of the declaim statement in my code. And I certainly didn't expect the sbcl team to analyze my code. Indeed, the reason for sending it was to see if anyone could reproduce the defect on other configurations.
So to be more specific, the declaim statement can be removed and the resulting behaviour is the same:
1. Crash on load, the first time, with debugger abort as per my original mail.
2. Successful load the second and subsequent times.
3. Crash on first exection of (doer 1) or (run 100 C-0), with debugger abort,
4. Successful execution ad infinum afterwards.
5. (compile-file "...") also fails the 1st time then succeeds on subsequent trials.
6. Loading the compiled version seems to work ok, but execution fails on the first go also.
That is about as specific as I can be as to the behaviour of the code in sbcl.
I run SBCL 0.8.19, on Ubuntu Hoary, which is Debian based.
As to my hw paltform:
AMD64 3000+ 939 pin
ASUS A8V Deluxe motherboard
Maxtor 160 GB SATA disk
1 GB kingston RAM
NEC ND3500 DVD-RW
ATI Radeon 9250 Graphics card.
All of this is spanking new, works well and shows no signs of hw problems.
That's the best I can do to describe the defect. I am happy to wait for the next version of SBCL to be able to access the debugger.