 Empirical Results
started 12/26/2000; 5:22:34 AM - last post 1/1/2001; 2:00:06 AM
|
|
Ehud Lamm - Empirical Results 
12/26/2000; 5:22:34 AM (reads: 467, responses: 2)
|
|
| Empirical Results |
|
Computer science is a mathematical science, and computer scientists prefer theorems to experiments. Alas, when it comes to studying the effect language features and other SE paradigms have in real life situations, empirical studies seem to be the most valid approach.
This site contains some interesting research. From our point of view the work on language features like inheritance and assertions (DbC) is most interesting.
By the way: When it comes to studying the effects of inheritance there are contradictory results in the literature. How deep is too deep?!
Posted to "" by Ehud Lamm on 12/26/00; 5:23:43 AM
|
|
|
|
water - Re: Empirical Results
12/31/2000; 5:19:09 PM (reads: 487, responses: 0)
|
|
|
Luca Cardelli's publications include a great book titled A Theory of Objects, which basically adds to his results in his published papers. He does some really systematic and useful analysis of the different possibilities in inheritance and delegation semantics in various styles of object-oriented languages. The most interesting result I found concerned covariance and contravariance over parameters in typing objects.
Anyway, the approach he focuses on (his particular theory of objects foundation) is to formalize objects as a kind of record of methods, with inheritance mechanisms being invisible to the theory. I found it an interesting starting point in reasoning about which mechanisms I preferred and how they affected user programming.
|
|
Ehud Lamm - Re: Empirical Results
1/1/2001; 2:00:06 AM (reads: 484, responses: 0)
|
|
|
Can you elaborate on how the thoery effected your views?
I am not sure I see the connecton between theory and actual use. I think actual use depends on factors outside the scope of the theory, like human psychology factors. Another example: in theory unbounded nesting is great; in practive it soon gets intractable. Same goes to inheritance, IMHO.
|
|
|
|