I do not like the syntax for std::expression. It violates the principle of least astonishment - anyone who sees (LHS) = (RHS) expects RHS to be evaluated. At the very least that assignment operator would need to be swapped out for something that does not apply evaluation, but I'd still prefer to step away from the DSL and have something along the lines of a std::invoke_later(func, params...) instead. 

If it is equivalent to the lambda as you suggest, what happens when it goes out of scope? Sure, those who are used to writing this in lambda form will know to use captures, but here that is not catered for. That's about all I have for the std::expression type as a standalone entity.

For the prologue and epilogue, the original ensures and expects syntax made sense. My problem with the proposal was the inflexibility and somewhat bizarre categorisation, rather than something simple like an unsigned integer debug level which can be set as a compilation flag (such that 0 only evaluates debug level 0, 1 evaluates 0 and 1, 2 evaluates 2,1 and 0, and -1 evaluates none of them), a binary mask, or a keyword lookup. The expects and ensures could also have been easily identified during optimisation steps to match up assurances with expectations. I think this is less clear. It is also unclear how the prologue can know the names of the input variables, given that they're defined afterwards in the standard function definition.

On Tue, 6 Aug 2019, 09:19 Andrew Tomazos via Std-Proposals, <std-proposals@lists.isocpp.org> wrote:
Please find attached DRAFT 1 of:

    Proposal of Contract Primitives

Feedback appreciated.

Std-Proposals mailing list