Re: Container Purity

2015-12-09 Thread Nordlöw via Digitalmars-d-learn
On Wednesday, 9 December 2015 at 10:47:18 UTC, Nordlöw wrote: that uses std.experimental.container Correction: I mean std.experimental.allocator

Container Purity

2015-12-09 Thread Nordlöw via Digitalmars-d-learn
Is it currently possible for a container such as https://github.com/economicmodeling/containers/blob/master/src/containers/dynamicarray.d that uses std.experimental.container to be completely pure? If not can be made to? I wonder because only length(), empty(), front() and back() are tagged

Re: Container Purity

2015-12-09 Thread Kagamin via Digitalmars-d-learn
Allocators usually use global state. Such code is usually treated as impure.

Re: Container Purity

2015-12-09 Thread Nordlöw via Digitalmars-d-learn
On Wednesday, 9 December 2015 at 11:46:45 UTC, Kagamin wrote: Allocators usually use global state. Such code is usually treated as impure. What about containers that store their own local allocator? Will DMD infer all members of such containers to be pure if they only access locally