Re: Gee Functional iterators - call for consensus



On Wed, 2011-07-06 at 19:41 +0100, Maciej Marcin Piechotka wrote:
> Sorry for late response.
> 
> To sum up the discussion:
> 
> 1. A common parent interface for the Iterator and Iterable which returns
> Iterator after calls to map etc. 
> 
> 2. Add a few more methods (slice). filter/slice require fixing bug in
> vala to provide implementation.
> 
> 3. State explicitly in documentation when does the function finish (I'd
> be grateful for any help here)
> 
> Open points:
> 
>  - Name of the parent interface. (Forallable? ;), Foldable?)
>  - Should it be implemented as virtual or abstract methods. As I stated
> I prefer C#/Haskell approach then Java and I'm inclined toward the
> virtual methods[1].
>  - Is there any other useful operations (tee?)
>  - Should tee/unfold go into parent interface or stay in Iterator
> 
> I hope I will have updated patches to share by the end of weekend. 
> 
> Regards
> 
> [1] In this particular case when the operation is clearly defined by the
> default implementation even if it is not necessary the most optimal one.

Large (+435-195) patch landed into my gitorious branch. Random
sidenotes:

 - I've named it Traversable. It might not be the best name but I'm
still waiting for better sugestion
 - The current functions are implemented. I'll add slice, filter.
Implementation of Iterator.tee would be also nice:

 ArrayList<Iterator<G>> tee(Iterator<G>, int n) { ... }

 - Example why the abstract classes would cause problems - currently we
use inheritance internally to share large portion of codes in iterators.
Moving to AbstractIterator would require to duplicate part of code.
 - If other maintainers agree (I've added them to CC) I would like to
push the changes to master before the end of Desktop Summit. 

Regards

Attachment: signature.asc
Description: This is a digitally signed message part



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]