druntime has a hash implementation in module rt.util.hash that looks pretty useful, but isn't publicaly exported by the library anywhere.

Does phobos or druntime offer a public hash function like that? If no, can it expose this one?

On Fri, Feb 15, 2013 at 03:26:28AM +0100, Adam D. Ruppe wrote:
> druntime has a hash implementation in module rt.util.hash that looks pretty useful, but isn't publicaly exported by the library anywhere.
> 
> Does phobos or druntime offer a public hash function like that? If no, can it expose this one?

+1. This is a very good and useful hash function that should be usable by user code. No need to force the user to reinvent the wheel or copy-n-paste code.


T

-- 
The diminished 7th chord is the most flexible and fear-instilling chord. Use it often, use it unsparingly, to subdue your listeners into submission!

On Friday, 15 February 2013 at 02:26:31 UTC, Adam D. Ruppe wrote:
> druntime has a hash implementation in module rt.util.hash that looks pretty useful, but isn't publicaly exported by the library anywhere.
>
> Does phobos or druntime offer a public hash function like that? If no, can it expose this one?

This might be a good time to bring up something that was bothering me when I examined the current druntime hash implementation. Just out of curiosity I was researching hash functions, and I found the hash function expert Bob Jenkins' FAQ about hashes:

http://burtleburtle.net/bob/hash/hashfaq.html

He says specifically that any hash function requiring a modulo prime at the end is bad (inefficient), but if I'm not mistaken, the druntime does just that. At line 50 of the associative array module there is a list of primes, and I believe the implementation assumes them to be necessary.

https://github.com/D-Programming-Language/druntime/blob/master/src/rt/aaA.d

I don't know enough to be sure this is a problem. There might be some other reason for the modulo prime, but I'd like to put my mind at ease on it.

On 2013-02-15 03:26, Adam D. Ruppe wrote:
> druntime has a hash implementation in module rt.util.hash that looks
> pretty useful, but isn't publicaly exported by the library anywhere.
>
> Does phobos or druntime offer a public hash function like that? If no,
> can it expose this one?

TypeInfo has a "getHash" method.

-- 
/Jacob Carlborg

On Fri, Feb 15, 2013 at 08:28:34AM +0100, Jacob Carlborg wrote:
> On 2013-02-15 03:26, Adam D. Ruppe wrote:
> >druntime has a hash implementation in module rt.util.hash that looks pretty useful, but isn't publicaly exported by the library anywhere.
> >
> >Does phobos or druntime offer a public hash function like that? If no, can it expose this one?
> 
> TypeInfo has a "getHash" method.
[...]

Which has a lot of bugs and inconsistencies. Like the fact that hashing
a char[], a const(char)[], and a string (immutable(char)[]) gives
different results.


T

-- 
Blunt statements really don't have a point.

On 2013-02-15 16:13, H. S. Teoh wrote:

> Which has a lot of bugs and inconsistencies. Like the fact that hashing
> a char[], a const(char)[], and a string (immutable(char)[]) gives
> different results.

Didn't know that. I just knew it existed.

-- 
/Jacob Carlborg

On Friday, 15 February 2013 at 03:33:13 UTC, Zach the Mystic wrote:
> This might be a good time to bring up something that was bothering me when I examined the current druntime hash implementation. Just out of curiosity I was researching hash functions, and I found the hash function expert Bob Jenkins' FAQ about hashes:
>
> http://burtleburtle.net/bob/hash/hashfaq.html
>
> He says specifically that any hash function requiring a modulo prime at the end is bad (inefficient), but if I'm not mistaken, the druntime does just that. At line 50 of the associative array module there is a list of primes, and I believe the implementation assumes them to be necessary.
>
> https://github.com/D-Programming-Language/druntime/blob/master/src/rt/aaA.d
>
> I don't know enough to be sure this is a problem. There might be some other reason for the modulo prime, but I'd like to put my mind at ease on it.

Without objection, so filed...

http://d.puremagic.com/issues/show_bug.cgi?id=9522