On 3/12/18, Peter Sommerlad <peter.sommerlad_at_[hidden]> wrote:
> Hi folks
>
> I took the task from MISRA C++ to write a paper on senseful class
> designs, i.e., categories of class types with respect to
> defining/declaring special member functions with the least amount of
> safety risks, i.e. by developer surprises. C++ syntax allows much more
> than is useful in that context, so a goal is to restrict (rule-of-zero
> vs. rule-of-plenty) and funnel useful construction principles for class
> types.
>
> Initially MISRA C++ came up with the following useful type categories:
>
> * value types (following rule-of-zero, copyable in general)
> * monomorphic object types (no-copying, no virtual, just encapsulation
> of (mutable) stuff to implement an abstraction)
> * polymorphic object types (no-copying, no-move, defaulted virtual dtor
> in base class)
> * resource management types (move only, or sharing, non-default dtor) -
> might be obsoleted by unique_resource and existing RAII types, but not
> sure in general (own value types might need that for memory, but I
> believe they could/should use unique_ptr then, but still might require a
> special dtor for deleting contained elements of a non-default
> constructible type)

Maybe add container type.

>
> There might be exceptions, but MISRA expects a paper to be able to refer to.
>
> I wonder in which of the plenty (and unfamiliar to me) sections of WG23
> paper such a guideline would fit, if at all. Having it as a WG21-SG12
> paper is OK as well, but I think WG23 could benefit from it as well.
>
> Further ideas and guidance appreciated. I hope to have such a paper
> ready for Rapperswil meeting (time permitting).
>
> Regards
> Peter.
> --
> Prof. Peter Sommerlad
>
> Institute for Software: Better Software - Simple, Faster!
> HSR Hochschule für Technik Rapperswil
> Oberseestr 10, Postfach 1475, CH-8640 Rapperswil
>
> http://ifs.hsr.ch http://cevelop.com http://linticator.com
> tel:+41 55 222 49 84 == mobile:+41 79 432 23 32
> fax:+41 55 222 46 29 == mailto:peter.sommerlad_at_[hidden]
>
> _______________________________________________
> ub mailing list
> ub_at_[hidden]
> http://www.open-std.org/mailman/listinfo/ub
>

-- 
Lawrence Crowl