bug#39586: pfds 0.3 hamt-fold crashes with guile-2.0 and guile-3.0

["Ludovic Courtès" <ludo@gnu.org>]

Hi Rob,

Rob Browning <rlb@defaultvalue.org> skribis:

> With the current pdfs (https://github.com/ijp/pfds
> 454033f82dac7c0b0ea9e84eed1e8ed316487c78), the code below halts on an
> rnrs record assertion failure with both guile-2.2 and guile-3.0, and I'm
> wondering if I'm doing something wrong, if it might be a guile bug, or
> if it's more likely an issue with pfds.
>
>   (use-modules
>    ((pfds hamts) #:prefix hamts/)
>    ((srfi srfi-69) #:prefix hash/))
>
>   (define x (hamts/hamt-set
>              (hamts/hamt-set
>               (hamts/hamt-set
>                (hamts/make-hamt hash/hash eqv?)
>                #:x 1)
>               #:y 2)
>              #:z 3))
>
>   (hamts/hamt-fold (lambda (k v result) result)
>
> It looks like the crash is in (rnrs records procedural) here:
>
>     (if (and parent (struct-ref parent rtd-index-sealed?))
> 	(r6rs-raise (make-assertion-violation)))

Do you think you could boil it down to a test that does not rely on pfds?

Thanks,
Ludo’.