On Tue, 25 Nov 2003 13:15:48 -0800 "Sean R. Lynch" <[EMAIL PROTECTED]> wrote: Though i'm not convinced that the all Python part must be rewritten in Pyrex. E.g. the pickling stuff is really complex and file loading is IMHO fast enough. Cloning / copying stuff enter in the same category.
Ok, that makes sense. People use Python for a reason, right? :) I was thinking that the Python stuff was mostly shadow classes, but I forgot that there were a lot of parts that aren't interfaces to C code at all.
Other modules like the facecutter are really specific and only used at design time, so i believe they don't worth conversion, at least in a first time.
Ok, I'll leave those too.
We must also take care of keeping REAL python class, so as we can extend them in pure python, and using mutiple inheritance and mixins.
Pyrex classes are Python classes, so this is no problem.
The problem will be to find your way in the C code, because it is quite a mess... What do you think of that Blam ?
Most of the C code will convert straight across with just a syntax change. Decoding the Python C API calls will probably be the hardest. It would also be nice if Pyrex would decode .h files, but that doesn't even work well with SWIG, so I should count my blessings :)
If you start any work, use the CVS version (which has lots of changes from the last release).
I have been using the CVS version for everything I've been doing.
I will probably have Land converted over within a couple of days, since that's the C code I've spent the most time looking at. The time is due to work and the holiday, not because of difficulty :)
The land.py file will end up looking like:
import _soya
Land = _soya.Land
:)
pgp00000.pgp
Description: PGP signature
