Croquet on mac os-9

Andreas.Raab at gmx.de Andreas.Raab at gmx.de
Sun Nov 17 20:56:30 UTC 2002


Hi John,

> Well keeping this stuff as a FFI call would be nicer. Once it gets into  
> a library, then the VM builders have to
> make new VM or plugins for fixes or changes & additions. If the call is  
> via FFI then any squeaker can fix it and distribute a change set.

Well in the case of OGL I'd say that there isn't a lot to change/fix since
the API is standardized (and nothing would prevent a quick hack which uses FFI
instead of the primitives). One of the reasons why I like the "static
bindings" here is that porting the FFI is much harder than just compiling a plugin
- I don't think that (for example) there are any ARM bindings for the FFI
yet. And considering the advances in handheld technology I think it's only a
matter of time when the first ARM machine will come with a *real* graphics chip.

> I think the only issue is that various platforms have different names  
> for the module. Usually the api name, parm types etc are the same.
> So can one substitute an instance variable here for example for that  
> hard coded 'OpenGLLibrary'? That for example is the only difference  
> between the OS9 and OSX version of the openGL interface class.

Yes, we could. I'd have to make the actual OGL bindings into a subclass of
ExternalLibrary which can figure out the name of the module based on the
platform we're running on. This wouldn't be hard to do. I'm uncertain though about
whether or not the calling conventions will be the same across platforms.

> Ah on the mac, stubs are statically linked in, then resolved to the  
> real address dynamically at runtime.

So no problems here. Good.

Cheers,
  - Andreas

-- 
+++ GMX - Mail, Messaging & more  http://www.gmx.net +++
NEU: Mit GMX ins Internet. Rund um die Uhr für 1 ct/ Min. surfen!




More information about the Squeak-dev mailing list