Iterators and Ranges: Comparing C++ to D to Rust

[Ola Fosheim Grøstad]

On Monday, 14 June 2021 at 13:30:52 UTC, Ali Çehreli wrote:
> popFront() consumes the range, not the container:

I glossed over the presentation, and I don't think he really 
covered things as well as he should. For instance there is a 
difference between generators, 
nonconsuming-iterators/table–pointers, consuming-iterators, but 
sometimes the distinction is blurry.

What if you create a range that consumes a queue. What if you 
create a range that interpolates and doubles the length of the 
source.

What if you want to modify the underlying container while 
iterating over it.

It would be nice with an ontology for table-pointers, iterators 
and generators, along the lines of the original Gang of Four book 
on design patterns, but I didn't really feel this presentation 
brought anything to the table. His take on this was very 
"syntactical".

Also, his assumptions about performance were not really well 
founded. You have to look at how iterators are used in actual 
alogrithms and how that could block optimizations to say anything 
sensible about that.

E.g. his take on StopIteration in Python is not really correct 
either, you can write a generator (coroutine) with no explicit 
StopIteration (although the semantics implies a raised exception, 
but that does not mean the implementation has to).