hash function

[Lars Tackmann]

Hi i need a hash function for some large strings (upto 1024 chars in the
array) the following is not very good for these large strings:

int string_hash(int hash_size, const char *key)
{
unsigned long n = 0;

const char *p;
for (p = key; *p != '\0'; p++)
n = ((n << 8) + n) ^ (unsigned long)(*p);
printf("n er: %i", n);
return (n % hash_size);
}

Any ideas -

 

[CBFalconer]

Hi i need a hash function for some large strings (upto 1024 chars in
the array) the following is not very good for these large strings:

hashlib.zip contains some string hashing utility functions,
together with some URL references to others. See:

<http://cbfalconer.home.att.net/download/>

 

[osmium]

Hi i need a hash function for some large strings (upto 1024 chars in the
array) the following is not very good for these large strings:

int string_hash(int hash_size, const char *key)
{
unsigned long n = 0;

const char *p;
for (p = key; *p != '\0'; p++)
n = ((n << 8) + n) ^ (unsigned long)(*p);
printf("n er: %i", n);
return (n % hash_size);
}

What is it that you don't like about what you already have? I have the
impression that MD5 is the _ne plus ultra_ of hash functions if you aren't
concerned about speed.

 

[glen herrmannsfeldt]

Hi i need a hash function for some large strings (upto 1024 chars in the
array) the following is not very good for these large strings:

int string_hash(int hash_size, const char *key)
{
unsigned long n = 0;

const char *p;
for (p = key; *p != '\0'; p++)
n = ((n << 8) + n) ^ (unsigned long)(*p);
printf("n er: %i", n);
return (n % hash_size);
}

I would use CRC32, and there is already a thread discussing it.

-- glen

 

[Paul Hsieh]

(e-mail address removed) says...

Hi i need a hash function for some large strings (upto 1024 chars in the
array) the following is not very good for these large strings:

int string_hash(int hash_size, const char *key)
{
unsigned long n = 0;

const char *p;
for (p = key; *p != '\0'; p++)
n = ((n << 8) + n) ^ (unsigned long)(*p);
printf("n er: %i", n);
return (n % hash_size);
}

I belive that you will find the state of the art in hash functions here:

http://www.isthe.com/chongo/tech/comp/fnv/

 

[August Derleth]

38:17a:

I have the
impression that MD5 is the _ne plus ultra_ of hash functions if you
aren't concerned about speed.

What about SHA? I think SHA is slowly replacing MD5 in most arenas.