 An Interview with A. Stepanov
started 9/8/2000; 6:10:40 PM - last post 10/3/2000; 4:49:30 AM
|
|
Chris Rathman - An Interview with A. Stepanov 
9/8/2000; 6:10:40 PM (reads: 1057, responses: 5)
|
|
| An Interview with A. Stepanov |
|
Stepanov is always an interesting read. He's especially critical of Object Oriented Programming, though I think he's being myopic in his vision of software. Algorithms are not applications, they are pieces of software that have to be screwed together in interesting ways. It's easy for a person who's sole responsible is to develop algorithms to forget that they are not the end of the story in software construction.
Beyond that, I wonder if Stepanov is familiar with ML, Haskell and other Functional Programming languages. These FP languages seem a much more natural fit to generic Type systems than the pointer finagling you have to do in C++ to get genericity and covariance. I can't say that I've ever been impressed with C++ as a language, or with the C++ Template mechanism.
That said, Stepanov's work is quite impressive and he deserves to be listened to.
Posted to "" by Chris Rathman on 9/8/00; 6:11:35 PM
|
|
|
|
Ehud Lamm - Re: An Interview with A. Stepanov
9/9/2000; 3:49:53 AM (reads: 1096, responses: 0)
|
|
|
|
Ehud Lamm - Re: An Interview with A. Stepanov
9/9/2000; 11:26:20 AM (reads: 1090, responses: 0)
|
|
|
For the record: I am a great fan of generic programming. Esp. the Ada approach.
I think the C++ template mechanism has some interesting features, notably automatic instantiation, and specialization. These two are absent in Ada. Comparing the two generic programming facilites can be illuminating. A good place to start thinking about these issues is the Ada-83 Rationale. The Stepanov interviews and papers, have plenty of good ideas too.
|
|
andrew cooke - Re: An Interview with A. Stepanov
9/11/2000; 10:03:43 AM (reads: 1079, responses: 2)
|
|
|
From what I understand, Stepanov has objections to OOP that go deeper than contravariant methods, but a lot of his arguments about the advantages of the STL (I know nothing about Ada generics, so I won't add them in at this point) over, say, Java seem to focus on the problem of returning specialised types.
So is there any record of his reaction to Eiffel? I just did a search for "Stepanov" and "Eiffel", but it didn't turn up anything interesting... (Eiffel has covariant methods). Or any evidence that Eiffel is particularly good at encapsulating algorithms?
|
|
Ehud Lamm - Re: An Interview with A. Stepanov
9/12/2000; 11:29:38 AM (reads: 1133, responses: 0)
|
|
|
Is it just me, or is it in general much easier to produce good generic code as opposed to good code that uses inheritance?
|
|
andrew cooke - Re: An Interview with A. Stepanov
10/3/2000; 4:49:30 AM (reads: 1141, responses: 0)
|
|
|
From what I understand, Stepanov has objections to OOP that go deeper than contravariant methods, but [...] So is there any record of his reaction to Eiffel?
Someone else thought similarly - well, they also knew a lot more about generics in Eiffel than me - see top item here
Andrew
|
|
|
|