[SciPy-dev] Inclusion of cython code in scipy

[Nathan Bell]

On Wed, Apr 23, 2008 at 10:28 AM, Gabriel Gellner <ggellner at uoguelph.ca> wrote:
> I am confused by what Cython disappearing means. It is pure python, and
>  generates C code.

So I can take an arbitrary .pyx file from Sage and use it sans-Cython?

> Lets say it is no longer maintained, I can still build my
>  existing code using the last version and I am no worse off.

Good luck with that.  I hope it's bug-free and feature-complete.

>  For my code to become useless python would also have to disappear
>  (or change so dramatically that I couldn't use my old codes). By this logic
>  no core tool should be written in python either?

See first point.

>  The code is open source it can't be taken away from me, it just might become
>  archaic, but that is no different from the current situation with g77, which
>  is still used extensively.

Bad analogy.  Fortran 77 still exists because a lot of people have
Fortran 77 code that's worth keeping.

In fact, the Fortran analogy works against you.  Consider something
like LAPACK which was written long before people cared about Python.
You can access LAPACK via a dozen different higher-level and
cross-language interfaces.  Why?  Because it's tied to a stable,
well-supported programming language and can be interfaced to other
languages/environments as necessary.

>  I agree that C++ is more portable, but are there examples of open tools
>  really disappearing?

Take a look at sourceforge.  While those projects haven't
"disappeared" they're not exactly useful either.


Am I really the only one who expresses the least bit of concern about
writing lot of Cython code?


-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/