Sam Steingold wrote:
> Raymond Toy wrote:
>> I have this application that I'm trying to debug. When a particular
>> function is compiled, I get an error message:
>> *** - SYSTEM::STORE: index 4003 for
>> #(0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
>> 0.0 0.0
> do you get this during compilation?
No, it's from running the test.
>> The array is clearly not 4000+ elements long. :bt says it occurred
>> in the routine o8opti.
> try looking at the disassembly of o8opti - does it look right?
I'd rather not. o8opti is about 2000 lines long. (No, I didn't write
it. It's an f2cl conversion of a Fortran routine that is some 1200
>> Ok. How can I figure out where the problem is?
> try splitting o8opti into smaller functions to figure what is trying to
> store where.
Possible, but tricky since it's originally one giant routine. :-(
>> To complicate things, I thought I could get better debug information
>> if I just loaded up the lisp file instead of compiling it. Well, then
>> my test runs to completion.
> looks like we are on a way to discovering a compiler bug.
> please try to create a small test case.
If I could, I would, but this isn't so easy.
I'm fairly confident that the translation is correct because cmucl and
ecl run the test correctly.