On Wed, 15 Oct 2025 at 00:22, Louis Dionne <ldionne.2_at_[hidden]> wrote:
> I do think your papers makes claims that are incorrect w.r.t. the usability of Contracts or their natural fit for being used with hardening. I disagree with those parts of the paper, I just chose not to highlight that in my reply.

Those parts of the paper are specifically talking about P2900/C++26
contracts. Those are a bad fit, and there's no way around that. There
are existing
scenarios where the choice of whether hardening is on needs to be and
needs to remain independent of the choice of a contract evaluation
semantic
for other code (meaning that stdlib hardening is on even if you choose
to ignore your own contract assertions). So, a TU-global toggle is a
bad fit for that.

That's not a claim.