[NTLK] Proof of Concept: Moving data from Einstein to iOS
jake at allaboutjake.com
Mon Nov 23 20:32:19 EST 2015
A full Objective-C bridge is more ambitious than I had considered, but an interesting idea that I'm going think about some more. I want to look around at other Objective-C bridge API's to see if theres a simple one that I can mimic.
I've cleaned things up a bit in my latest tests. I added a new "coprocessor" to the emulator, essentially. NewtonScript calls a function which triggers an MCR instruction similar to the way you did the other native calls. Works fine and keeps the libffi code separate and untouched.
I'll check in a branch after a little more work, in case people will find it useful, along with a newton PKG that implements the transport.
On 11/23/15, 3:36 PM, "newtontalk-bounces at newtontalk.net on behalf of Paul Guyot" <newtontalk-bounces at newtontalk.net on behalf of pguyot at kallisys.net> wrote:
>> Building REX from scratch is quite difficult as you need the MacOS toolchain. I have written the "mosrun" interpreter ("MacOS runtime environment") that lets you run all the Newton MPW tools on OS X (and probably any other OS, as the code is plain C++).
>> I did manage to compile the network driver module for Einstein and link it into the REX, but I was not able to build a new Rex from scratch. Paul code is missing resource forks, and I don't remember if he was able to find them in 2013 when we tried all this.
>I don't remember having issues with resource forks, yet I do remember using mosrun to build a new REX not so long ago (was it in 2013 ?). But I might have forgotten.
>libffi definitely doesn't work on 64 bits systems : even if you could get support for iOS it is completely broken.
>However, a clean way to handle what you try to achieve would be to implement an ObjC-NewtonScript bridge for iOS and MacOS X, and this would definitely replace libffi on these platforms.
More information about the NewtonTalk