Re: [orca-list] Draft Proposed Two-Year Roadmap

[Jason White <jason jasonjgw net>]

Trevor Saunders <trev saunders gmail com> wrote:
 
> * Determine the performance benefits, if any, of re-writing Orcaâs core
> * code in C/C++.*
>
> I'll be really supprised if it doesn't help a lot, I've seen a couple
> places where all python and c / python speed has been compared in the
> real world, and c / python is generally a lot faster, even when almost
> all the time should be spent in system calls so the speed really should
> be basically the same.

Two other possibilities come to mind. I don't know whether these have been
investigated and dismissed already, however.

1. Pypy, which, when complete, should out-perform the current Python
interpreter significantly.

2. Vala, which is already a gnome project. It's an object-oriented programming
language with built-in memory management that can be compiled into C code.

My concern regarding C and C++ is the risk of segmentation faults due to
pointer and memory management bugs, which seem to creep into C/C++ software
despite everybody's best efforts, especially with multi-threaded systems.