Alo, I'm wondering how phobos devs view or determine what goes in to std.algorithm and what goes in to std.range.

To me some of them are quite obvious - well, most things can arguably be an algorithm. But for example "refRange" is clearly a range specific thing, but "transpose" is not. And things that create ranges from nothing also may "clearly" belong in the range module? (e.g. iota, generate and recurrence)

Also curious, are there any github PRs in phobos where certain algorithms were discussed as going in to where and what the reasonings were?

cheers
- Ali

On 7/14/18 5:32 PM, aliak wrote:
> Alo, I'm wondering how phobos devs view or determine what goes in to std.algorithm and what goes in to std.range.
> 
> To me some of them are quite obvious - well, most things can arguably be an algorithm. But for example "refRange" is clearly a range specific thing, but "transpose" is not. And things that create ranges from nothing also may "clearly" belong in the range module? (e.g. iota, generate and recurrence)
> 
> Also curious, are there any github PRs in phobos where certain algorithms were discussed as going in to where and what the reasonings were?
> 
> cheers
> - Ali

There is no hard and fast rule. If it does something one would naturally associate with an algorithm (e.g. it would belong in a book on algorithms etc), it belongs in std.algorithm even if it processes ranges. If something is algorithmically trivial but support ranges, then it belongs in std.range.

Without looking: where should "chain" belong? It's a trivial algorithm but has a variety of intricacies for supporting different kinds of ranges. So it's in std.range.


Andrei