Lambda the Ultimate

Via Scott Rosenberg (whose book Dreaming in Code I think we mentioned here before), I learn than Susan Lammers is making the interviews from her 1984 book Programmers At Work available on the Web.

Here is how she describes the goals of her new site:

Many people have urged me over the years to do a second Programmers At Work (PAW) with a new generation of programmers and I’ve sketched out the project, made lists of new folks to feature, done inquiries, thought about going back to talk to the guys in the original edition, and other variations . Now with this web site, I will make the original interviews available online, and perhaps it will become the seed for a more “thoroughly modern” approach to the PAW series. What I’m hoping we can kindle on this site is an ongoing exploration and dynamic conversation with the "connected" community of programmers on the web about the creative process in programming.

Who better than the LtU community to contribute to such a conversation?

The first PAW interview posted to the site is the interview with Charles Simonyi, a man whose views of the future of programming, and programming languages, are mentioned here often (though not always with great enthusiasm).

Hi,

I am not sure I am entirely happy with the usual structural recursion on sets approaches. They all look like a bit of cheating, i.e. using lists under covers, broadly speaking.

I mean SRI/SRU/powerset/etc tricks (http://citeseer.ist.psu.edu/722866.html).

E.g. in the SRI method (collection insertion representation), we can do matching using the insertion operator that 'disassembles' a non-empty set thereby maing it amenable to something like foldr.

On the other hand, the 'choose' operator is criticised (elsewhere) as 'non-deterministic'. I am aware of various laws and such that make SRI 'work', I just have a feeling that the insertion operator is a kind of deus ex machina as much as 'choose' is !

Please comment.

Thanks.

Can someone confirm, refute, comment on, or point to the literature on my (somewhat inchoate) intuition that the access-control rules of the Bell-LaPadula model (specifically that a subject cannot read from objects of higher privilege, the "simple security property", and that a subject cannot write to objects of lower privilege, the "*-property") are just special cases of the safety of covariant results and contravariant arguments respectively?