Dfyjkt

I know that hash codes are generally the fastest way to check dynamic sets, but I was wondering what is the fastest way to check whether a dynamic string is in a read-only string set known at compile-time. (I mean mainly length: usize; chars: &[u8] strings, not ropes or cons strings.)

Currently, I'm usually doing stuff like this, but it seems like it'd be suboptimal:

// What I mean
let keywords = Set::new(["do", "if", "in", "for", "new", "try"]);
fun is_keyword(s: &str) keywords.contains(s) 

// What I write
function is_keyword(s: &str) 
 match s.length() 
 2 -> s == "do" 

Is there anything faster than something derived from this second variant for sets of C-style strings? Or is it as fast as I could reasonably get?

This is language-agnostic - I don't care what languages answers use. I'm just using Rust due to familiarity.

For a static set, you can use perfect hashing. This is essentially a hash table, but the hash function guarantees that every string in the set hashes to a unique index in the table.

To test a dynamic string, you just hash it to an index using the perfect hash function, and then see if the one and only string at that index matches the test string.

A google search will find you lots of different ways to do perfect hashing. One of my favorites is described here: http://cmph.sourceforge.net/papers/chm92.pdf

It's often used for keyword matching in compilers, or implementing switch/case on strings in languages that support that.

Like you said, seems like the fastest way would be to hash the strings. Your current way would take O(N) time to search for the largest string in the set, or for a string that isn't in the set at all.