Re: PyObjC on OS X on Intel
Re: PyObjC on OS X on Intel
- Subject: Re: PyObjC on OS X on Intel
- From: Ronald Oussoren <email@hidden>
- Date: Fri, 1 Jul 2005 14:22:19 +0200
On 7-jun-2005, at 15:09, Nicko van Someren wrote:
On 7 Jun 2005, at 12:51, Luc Heinrich wrote:
On 7 juin 05, at 13:24, Nicko van Someren wrote:
Presumably it should all "just work". PyObjC does not contain
any assembly code that I'm aware of and the underlying Python is
endian independent, so at least in theory it should just compile
as a fat binary and run.
"Python on Mac OS X for Intel is not going to be a seamless
transition."
<http://bob.pythonmac.org/archives/2005/06/06/python-on-mac-os-x-x86>
OK, I take back the comment about it "just work"ing. I didn't
realise that PyObjC was such a mess architecturally.
I'm a very long way behind w.r.t. reading mailinglists, but: what
mess? You need assembly code or a compiler at runtime when you want
to intercept arbitrary methods at runtime, and the same is true for
calling arbitrary methods or functions. Before we used assembly code
we had a very large number of C functions that could be inserted into
a class and would forward the call to a python function (basically
one for every supported method signature). The current architecture
is a lot nicer.
For anyone who didn't notice this yet: I ported PyObjC you OSX/Intel
and did so in a couple of hours at WWDC. This code is in the PyObjC
repository.
And w.r.t. creating universal binaries for python itself: doing that
in a fully generic way will be (a lot) more work, mostly because of
autoconf. But I expect that Python 2.5 will cleanly support universal
binaries out of the box, I'll work on that if I can get my hands on a
development system.
Ronald
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden