So I'm working away at getting Pyd to automatically wrap a class's operator overloads, and it's going great. Until, that is, I get to opApply.

Oh yeah.

Python does iterators differently. First, you call PyObject_GetIter on an object, which returns an iterator object. Then you just keep calling PyIter_Next on the iterator object until it returns null.

It turns out that wrapping opApply with this sort of interface is essentially impossible. Long story short, it requires static closures in D.

More specifically, it requires functionality equivalent to Python's "yield" keyword. This is a keyword that causes a function to act like a generator. As the generator is iterated over, it just keeps running the function, using the results of the yields as the iterator values. This is a classic Python way of creating an iterator for a class.

So, in the meantime, wrapped opApply functions will have to be more literally translated as functions with a callback argument.

-- 
Kirk McDonald
Pyd: Wrapping Python with D
http://dsource.org/projects/pyd/wiki

In article <e8akf8$p17$1@digitaldaemon.com>, Kirk McDonald says...
>
>So I'm working away at getting Pyd to automatically wrap a class's operator overloads, and it's going great. Until, that is, I get to opApply.
>
>Oh yeah.
>
>Python does iterators differently. First, you call PyObject_GetIter on an object, which returns an iterator object. Then you just keep calling PyIter_Next on the iterator object until it returns null.
>
>It turns out that wrapping opApply with this sort of interface is essentially impossible. Long story short, it requires static closures in D.
>
>More specifically, it requires functionality equivalent to Python's "yield" keyword. This is a keyword that causes a function to act like a generator. As the generator is iterated over, it just keeps running the function, using the results of the yields as the iterator values. This is a classic Python way of creating an iterator for a class.
>
>So, in the meantime, wrapped opApply functions will have to be more literally translated as functions with a callback argument.
>
>-- 
>Kirk McDonald
>Pyd: Wrapping Python with D
>http://dsource.org/projects/pyd/wiki

It seems that you need something like StackThread: http://www.digitalmars.com/drn-bin/wwwnews?digitalmars.D/39435 http://assertfalse.com

From what I have understood, it implements coroutines, e.g. generator like objects for D. It seems to still have some issues, but should be usable.