Re: [Axiom-developer] domains with parameters versus Python classes

[Gabriel Dos Reis]

"Ondrej Certik" <address@hidden> writes:

| > I don't know which C++ you're talking about, but the one that I use on
| > daily basis and work on has lists (std::list), dictionaries (std::map,
| > std::multimap, etc.), has many garbage collectors as libraries with
| > a popular one freely available from Hans Boehm's web site -- industrial
| > strengh.  You don't have to use pointers in C++ if you don't need to.
| > Many  languages claim they don't have pointers, but represent
| > everything as pointers.  You can have everything dynamic in C++ if
| > that really is hat you want.  And, by the way it has a library for
| > easly interfacing with Python -- check out Boost.Python from Boost.
| >
| > Let's our ignorance not interfer with reality.
| 
| I also use C++ frequently, Boot.Python is very, very slow to compile,
| better is to use SWIG. Hans Boehm garbage collector is of course a
| nice thing, but the STL lists, map and stuff is very cumbersome to use
| compared to python.

That is a remarkably different statement that the ones you made earlier.

  # Well, at least Python and Ruby are much higher level than C++ (there
  # are lists, dictionaries, garbage collector, no pointers, it's dynamic,
  # etc.). And also many times slower.

I should have guessed.

| Compare for example GiNaC and SymPy's
| implementation of the same algorithms - SymPy is like 1/2 (or even
| more) smaller, because C++ is very, very verbose.

Now, you're elevating the properties of a particular library to that
of a language and language library.  I'll refrain from qualifying that
process and leave it at that.

-- Gaby