[boost] Re: [review results] hash functions accepted

From: Thorsten Ottosen (nesotto_at_[hidden])
Date: 2005-03-22 07:39:42

"Daniel James" <daniel_at_[hidden]> wrote in message 
news:d1nff6$ph6$1_at_sea.gmane.org...
Thorsten Ottosen wrote:

> that Joachin don't like the idea is ok, it's just not an arguemnt in favor 
> of
> much.

I was just recapping what conversation there was. But it does make three
people opposed to the idea.
>>>>>>>>>>>>>>>

all decision should, if possible, be based on arguments not on personal 
opinions

> What are the obstacles? The unordered containers can define their
> own version.

But there could be other containers which also cause problems which we
can't anticipate. Or other types which have a sequence, but don't define
equality in terms of that sequence.
>>>>>>>>>>>>>>>

The rule of thumb is that you must make sure objects comparing equal
has the same hash-value; what the hash-value is is not a problem AFAICT.

> Btw, would you please describe why the unordered containers would "break"?

It's possible for two unordered containers to contain exactly the same
elements but in a different order. So they would get different hash
values for hash_range, but are considered equal (although equality isn't
usually defined anyway).
>>>>>>>>>>>>>>>>

if equality isn't defined, then it makes little sense to use them as index in 
a hash-based
container. In fact, it won't compile.

-Thorsten