Derek's hashmap was faster than than Go 1.0's built in hashmap. With Go 1.1 or 1.1.1 (don't remember which one exactly), the standard one did become faster, though Derek had decided to stick with it as is. I can ask him what the reason was... I don't recall right now.
I think I may know why they are sticking with the self-built version: random key selection. The built-in hash map doesn't include a method to discover a random key, but the gnatsd hashmap does.
Comments
Derek's hashmap was faster than than Go 1.0's built in hashmap. With Go 1.1 or 1.1.1 (don't remember which one exactly), the standard one did become faster, though Derek had decided to stick with it as is. I can ask him what the reason was... I don't recall right now.
I think I may know why they are sticking with the self-built version: random key selection. The built-in hash map doesn't include a method to discover a random key, but the gnatsd hashmap does.