Re: [boost] [rfc] Preliminary Hash Library

Subject: Re: [boost] [rfc] Preliminary Hash Library
From: Daniel James (dnljms_at_[hidden])
Date: 2010-05-08 10:30:07

On 7 May 2010 23:37, Scott McMurray <me22.ca+boost_at_[hidden]> wrote:
> On 6 May 2010 06:01, Daniel James <dnljms_at_[hidden]> wrote:
>>
>> This will collide with the existing 'boost::hash' (an implementation
>> of 'std::hash').
>>
>
> Good point.  unordered_map's hash definitely has stronger claim.

Only because it came first. If we were making the choice today,
something like 'boost::container::hash' would be fine. I'm now
wondering if it might be a good idea to move it - I think it's rarely
used directly, so most wouldn't even notice. How long should the gap
be between moving 'boost::hash', and allowing something else to use
the name? I think it'd have to be a least a year.

> The simplest fix would be just changing the namespace to
> boost::hashes.  Alternatively, the library could be renamed to
> Cryptographic_Hash and use boost::cryptographic_hash as its namespace,
> though that would cut off the possibility of following Paul Bristow's
> suggestion.
>
> Any preferences or other ideas?

I dislike plural namespaces as they can be confusing (I can never
remember when to use 'iostream' vs. 'iostreams'), but I can't think of
anything better.

Daniel