Re: need for macros (was Re: Icon)

[Michael Vanier <address@hidden>]

> Date: Sat, 22 Dec 2001 14:03:15 -0500 (EST)
> From: Shriram Krishnamurthi <sk@cs.brown.edu>
> 
> Anton van Straaten wrote:
> 
> > But assuming a goal of creating a maximally powerful language for "do not
> > try at home, use at own risk" scenarios, is there any reason aside from
> > performance to avoid first class values?
> 
> Yes: static verifiability.  First-classness is generally a direct
> enemy of verifiability, especially once you get past very low-level
> verification systems such as (traditional) types.  It's the constant
> trade-off that us verification people who also want to express systems 
> in a reasonable notation have to wrestle with.
> 
> Shriram
> 

Wow, this is an interesting statement.  Have you got any pointers to papers
that discuss this in more detail, or would you be willing to expand on
this?

Mike