Why UUIDs are Secretly Incredibly Fascinating

Part of the DevOops: Learning from Failure specialist track

UUIDs are used all over the place, but there are a ton of variants. They're theoretically absolutely unique, and yet dozens of companies and projects abandoned the standard to invent their own identifier schemas instead.

Why is that? and why does someone think they're interesting enough to talk about at a conference?

Let's go down this rabbit whole together, today you'll learn:

• why there were five versions of UUIDs in the original spec
• which versions are still useful today, and when you should consider using them
• why, even so, they weren't considered fit-for-purpose for many situations
• what's being done to fix that
• ~~what's the difference between UUID and GUID?~~

I'll bet you'll find the answers more interesting than you expected to!

Tom Eastman He/His • @tveastman@cloudisland.nz • @tveastman

In 2001 Tom handed in a programming assignment for a university class that came out to about two thousand lines of Java. His professor later shared their model answer to the problem, it was thirty lines of Python. Tom switched sides on the spot.

Tom is a senior software engineer for Kraken Technologies, and is the president of Python New Zealand, the charity promoting and supporting the Python language community in New Zealand.