Re: Pre-PPC: class :abstract

* Ovid <[email protected]> [2025-12-20 16:24]:
> On Sat, Dec 20, 2025 at 5:04 AM Darren Duncan <[email protected]> wrote:
> > That problem is avoided easily enough by the abstract class having
> > a default implementation, which in likely most cases would be
> > reasonable.
>
> That is a very dangerous assumption. The following is a trivial example:
>
> use v5.40.0;
>
> class StorageHandler {
>     method connect() {
>         say "Connecting to generic backend...";
>     }
>
>     # If a subclass forgets to override this, the app thinks data
>     # is being deleted when it isn't. This violates the "fail-fast" principle.
>     method delete_all_logs($user_id) {
>         # Doing nothing here is dangerous.
>         return 1;
>     }
> }
>
> class CloudStorage :isa(StorageHandler) {
>     method connect() {
>         say "Connecting to S3...";
>     }
>     # Developer forgot to implement delete_all_logs!
> }
>
> my $storage = CloudStorage->new;
> $storage->connect;
>
> # The dev thinks they are clearing sensitive data:
> $storage->delete_all_logs($user_id);
> say "Cleanup complete!"; # Oops

All I can see here is a straw man argument. What Darren said is
“a reasonable default implementation” but what you chose as an
illustration of your argument against it is a stub method, which is
not the same thing anyway – and one which violates its own required
properties at that. The problem with this example code is not even the
provision of a stub but the use of `return 1` as its method body rather
than `...` or a `die "Not implemented"` or some such (which would yield
exactly the failure that, as per your comment, is required by the fact
that the non-implementation of the method results in no data actually
being deleted). I see no argument against the concept of a reasonable
default implementation here.