I've noticed that some ranges in Phobos don't implement "opSlice(/**/)", when they provide "opSlice(indexLow, indexHigh)".

For example, algorithm.map, or Array.Range don't provide it.

I was wondering if this was just an oversight of the implementation, or if there was a special reason to choose not to provide it.

I've taken to writing: "writeln(stuff[])" whenever I want to print a collection of things, to make sure I don't just print the name of the "stuff" class. I don't stop and wonder if stuff is a range or not, and just append it []. I was surprised it did not work consistently.

...Or is there a reason I should stop writing "stuff[]"?

On Tuesday, July 10, 2012 23:35:02 monarch_dodra wrote:
> I've noticed that some ranges in Phobos don't implement
> "opSlice(/**/)", when they provide "opSlice(indexLow, indexHigh)".
> 
> For example, algorithm.map, or Array.Range don't provide it.
> 
> I was wondering if this was just an oversight of the implementation, or if there was a special reason to choose not to provide it.
> 
> I've taken to writing: "writeln(stuff[])" whenever I want to print a collection of things, to make sure I don't just print the name of the "stuff" class. I don't stop and wonder if stuff is a range or not, and just append it []. I was surprised it did not work consistently.
> 
> ...Or is there a reason I should stop writing "stuff[]"?

1. In general, I wouldn't recommend not caring whether a variable is a container or a range. It matters a great deal to the semantics of what you're doing.

2. At this point, [] is primarily used for getting ranges from containers, for slicing static arrays, and for array copy operations.

a[] = b[];

You can't depend on any range defining it. None of the isXRange templates require it.

That being said, there's no reason why at least some ranges couldn't define it. However, it would probably have to be restricted to forward ranges (making it the same as save), so you still wouldn't be able to just assume that [] would work. And unless we add it to what isForwardRange requires, you couldn't even rely on forward ranges having it, since it's not required of them.

- Jonathan M Davis

On Tuesday, 10 July 2012 at 21:51:44 UTC, Jonathan M Davis wrote:
> On Tuesday, July 10, 2012 23:35:02 monarch_dodra wrote:
>
> 1. In general, I wouldn't recommend not caring whether a variable is a
> container or a range. It matters a great deal to the semantics of what you're
> doing.
>
> 2. At this point, [] is primarily used for getting ranges from containers, for
> slicing static arrays, and for array copy operations.
>
> a[] = b[];
>
> You can't depend on any range defining it. None of the isXRange templates
> require it.
>
> That being said, there's no reason why at least some ranges couldn't define it.
> However, it would probably have to be restricted to forward ranges (making it
> the same as save), so you still wouldn't be able to just assume that [] would
> work. And unless we add it to what isForwardRange requires, you couldn't even
> rely on forward ranges having it, since it's not required of them.
>
> - Jonathan M Davis

Thank you for your reply, it makes a lot of sense.