Python training time (was)

[Andy Freeman]

"Brandon Van Every" <vanevery at 3DProgrammer.com> wrote in message news:<Ym20a.492$1q2.48266 at newsread2.prod.itd.earthlink.net>...

> I'd estimate modern C++ at a 2 year learning curve to know every aspect of
> the language.  In 1994 it was a 1 year learning curve.

That estimate seems to be way off for any useful definition of "know every
aspect".

For example, something keeps Van Every from using templates for simple
reuse in the same way that he uses procedures.

> Andy Freeman wrote:
> > unless, of course,
> > you want to argue that C++ runs slower than C++.  (Hint: Martelli
> > described one technique for developing in Python and delivering C++
> > that is faster than developing in C++ and I described another.)
> 
> I'm not going to buy that writing in Python then converting to C++ is faster
> than writing in C++ in the first place for fairly simple problems.

"A simple problem" isn't enough to keep C++ competitive.  The problem
has to have a simple, small, and known solution to keep up.  Python
gets folks to known faster.

And, as far as conversion speed goes, generators put out >10k
lines/minute.  For our purposes, that's instantaneous.  A
generator for a huge fraction of an application is a small bit of
programming.  Auto generating the whole application is more work,
but I can switch to manual conversion at any point.

John Henry lost.