pytrin wrote:This is why it's important to read the posts as part of a complete conversation. I was responding to this cristiano fellow who suggested using custom functions to generate strings from the auto-incrementing identifier and I simply suggested that there hashing functions might be better used instead of creating a custom function with an unknown (probably worse) collision matrix.
I wasn't trying to start a debate on the merits of hashing functions for such purposes versus their theoretical definition.
I did follow the debate and nobody except you has mentioned hashing functions
Also, I still can't agree "a standard hash function" is better to use than a custom one.
First, because it has collisions for sure (and indeed, for new hashing functions it's an "unknown collision matrix").
Second, a 1:1 function is not "a custom function with an unknown (probably worse) collision matrix" - simply, it has no collisions.
So, applying a 1:1 custom defined function to an autoincrement value is much, much better than using a hashing function.