Wikipedia talk:Algorithms on Wikipedia/More talk

From Wikipedia, the free encyclopedia

A load of older talk can be found at Talk:Algorithms on Wikipedia

A couple of points:

  • OOP is about programming in the large. For programming in the small (to provide illustrative examples of algorithms, for instance), OOP is just a bunch of unnecessary cruft. Hence, Java and C++ don't add very much over C for the purpose of illustrating fundamental algorithms, and the cruft gets in the way of understanding.
  • Some of the HLL's (Perl, for instance), aren't exactly ideal for describing implementations of data structures (trees, hash tables etc).
  • As far as the untestability of pseudocode, most algorithms we present should be verifiable by inspection.
  • Perl looks ugly and encourages, er, idiosyncratic coding. It's great for getting jobs done. It's bad for presenting examples in, IMHO.
  • Of course, I think we should present most of our algorithms in something like Haskell, but I can't see us winning that argument :) --Robert Merkel
Yeah Robert I agree at most. Especially for C as I don't understand all the HLL's, PERL's, IMHO's, Haskell 98's, Haskell++, O'Haskell's and Mondrian's stuff you've written. C++ and Java are very active nowadays and perhaps according to C++, through C# and some other futher classification efforts C will see its own rebirth. I am also very much interested in RPN programming and in Maple V R4.00a codeing for the number theory applications. --XJamC 4 Wednesday (Thor's day) [2002.02.28) (0)

I suggest as a compromise the Python programming language one of the most readable progamming languages, often described as "executable pseudocode". -- Anon.

I would be very much in favour of real pseudo-code. This has benefits that

  • No programmer is "angry" because the implementation is in a - in his eyes - stupid, silly, inferior programming language (it's NPOV). Moreover, they will not feel compelled to add other versions of the same algorithm
  • No details of implementation have to be presented, and use of mathematical signs (also variables with subscripts) or plain text make the code very readable
  • Pseudocode makes it possible to focus on what's important about the algorithm.
  • An encyclopedia is about things, not the things themselves. Therefore, having executable code is not very useful (it's not in all languages anyway). If somebody is interested in implementing one of the algorithms, pseudocode should give him more than enough information to implement things in his favourite language, especially if the article also contains some pointers on implementation details (f.e., an algorithm on graphs can point to an article Graph representation in programming, or something like that.

I could probably go on here, but I think my point is clear. A problem of pseudo-code is of course that there's no standard, but it should not be a problem to create a Wikipedia pseudo code language, and linking to explanation when used. Jeronimo


If you fully define a pseudocode then it becomes a code. Next thing someone will write a complier for it, then people will start with the language pissing matches etc.... ;-)
Hehe, you might be right. But te definition should simply define some syntax (assignment, if-then-else, while-do, etc). The rest of the notation should be relatively free. That's why it's pseudo-code. Jeronimo