[cells-devel] Re: cells test breakloop

[Kenny Tilton]

mikel evins wrote:

> Kenny,
>
> When running cells::cv-test I hit a breakloop; the cells docs say that 
> this means something's wrong.
>
> Just for a sanity check (and because I'm considering whether to use 
> cells for some knowledge modeling in a personal project) I tried 
> running the tests from Lispworks, and broke in exactly the same way. 
> The output below is copied from OpenMCL 0.14.1, but is exactly the 
> same (modulo object numbers and break prompts) as that from Lispworks: 

Ah, that's one I had fixed for a while and then for some reason decided 
to break again. I am fixing it over here, but you can just either:

1. move on to try lesson-14 or starter; or

2. In c-break and c-calculate-and-set, change BREAK to ERROR

Lispworks is right, ACL is wrong: break should land you in the debugger 
regardless of any errors. What I am doing in a few places during CV-TEST 
is deliberately coding usage errors with Cells to confirm user errors 
are signalled, but then I want the test to run to completion, so I trap 
the errors and actually signal a "test-failed" error if no error gets 
trapped. Under ACL the /real/ Cell error gets signalled with break and 
the test can trap it, but that is wrong.

kenny