On Sun, Mar 2, 2014 at 12:16 AM, Roger Hui <[email protected]>wrote:
> See also *A Letter from Dijkstra on APL > <http://archive.vector.org.uk/art10501260> .* I get the distinct impression that Dijkstra's objection was that APL's interface was - at least for that time - far too easy to use. I could also argue that Dijkstra was far more interested in the cult of Dijkstra than the cult of Iverson, and not entirely without justice. But, looking deeper, I think what he was really objecting to was a professor making reference to a media type which he himself did not have access to. One expects a professor to either present the idea clearly or present a reference that can be accessed. When the reference is closed away and inaccessible an academic will traditionally object to the form of the presentation. (Or at least, an academic who prizes formalisms - and I think it's clear that Dijkstra had a fondness for formalism.) For me, though, I am particularly amused by the adoption of Python as the basis for computing curriculum. Mind you, I think it was a great choice, but more for the community behind it than the language itself. We have a lot to learn, from that community (and other communities as well). Still, out of respect for Dijkstra and his concepts of teaching computer science - I expect that we should more often be using Iverson's notation in expositions of interesting ideas. And that's where I think Python's "only one way to do it" quip falls down. A formal proof expressed as code involves showing a variety of ways of expressing the same concept, with easily describable manipulations between each expression. Of course, this immediately leads to systems dedicated to the implementation of particular flavors of formal proof (Coq comes to mind - and dependent typing can be thought of as sort of formal expression of the concept of an interpreter). But it's going to be up to us to present those ideas, and that's going to involve some compromises between the failings of formal systems and the failings of industrial practices (and the joys of each). Thanks, -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
