Re: [boost] typelists: MPL tutorial + partial review

From: Andrei Alexandrescu (andrewalex_at_[hidden])
Date: 2001-12-14 16:11:45

From: <Iain.Hanson_at_[hidden]>
> Author:  andrewalex (andrewalex_at_[hidden]) at unix,mime
> Date:    13/12/01 21:32
> >I think a compelling MPL example would be one that:
> >
> >1. Solves a real-world problem.
> >2. Cannot be solved simpler without basically rewriting a significant
subset
> >of MPL.
> >3. Is comprehensible outside a handful of people.
>
> I don't personally consider 3 to be that important.

I mentioned under what caveats 3 applies. FWIW, there are many LISP idioms
that have remained mysterious to anyone but a small group of experts.

> Alexy posted a rationale for mpl after you stated that it was to complex.
I
> have seen you address that yet.

That was because I kept my opinion, so there was nothing I'd add.

> If you have a solution to the GenScatter that is simpler and equally
> generic the please post it. Assertions don't tend to carry much technical
> information.

The original implementations of GenScatterHierarchy and GenLinearHierarchy
are to be found at http://moderncppdesign.com, and the explanations are in
MC++D. Also, AbstractFactory and Tuple in Loki use the two generators. I
could not use it in Visitor because of a bug in MWCW, which forced me to
expand the hierarchy by hand.

Andrei