Hi,

Han-Wen Nienhuys <hanwen@xs4all.nl> writes:

> Ludovic Courtès escreveu:
>
>> Not quite actually: the "hell = scm_malloc (...)" bit is still broken.
>
> ?

I fixed it, added a ChangeLog and NEWS entry and a test case, and pushed
it to 1.8.

The simplest way to trigger a `go_to_hell ()' call is this:

  (define-class <foo> (<object>) (the-slot #:init-keyword #:value))
  (define f (make <foo> #:value 2))
  (define-class <foo> (<object>) (the-other-slot) (the-slot))
  (slot-ref f 'the-slot)   ;; -> via `TEST_CHANGE_CLASS ()'

The test case is a variation on this, to make it likely to be hit by
out-of-bound accesses to HELL.

Thanks!

Ludo'.