I love reading about the normal forms, because it makes me sound like I know what I'm talking about in the conversation where the backend folks tell me, "if we normalized that data then the database would go down". This is usually followed by arguments over UUID versions for some reason.

▲

necovek 7 hours ago | parent [-]

So which normal form do they argue for and against? And what UUID version wins the argument?

▲

tadfisher 6 hours ago | parent | next [-]

Explaining jokes is poor form.

	▲	necovek 4 hours ago \| parent \| next [-]
		This was an attempt to extend jokes and not ask for explanation: there are a number of normal forms, and people usually talk about "normalization" without being specific thus conflating all of them; out of 7 UUID versions, only 2 generally make sense for use today depending on whether you need time-incrementing version or not.
	▲	culi 6 hours ago \| parent \| prev [-]
		On the internet it is normal.

▲

Tostino 6 hours ago | parent | prev [-]

Not OP, but UUID v7 is what you want for most database workloads (other than something like Spanner)

▲

tossandthrow 4 hours ago | parent | next [-]

I use the null uuid as primary key - never had any DB scaling issues.

▲

petalmind 4 hours ago | parent [-]

Yeah, no NULL is ever equal to any other NULL, so they are basically unique.

	▲	Groxx 4 hours ago \| parent [-]
		You are also guaranteed to be able to retrieve your data, just query for '... is null'. No complicated logic needed!

▲

RedShift1 4 hours ago | parent | prev [-]

Me still using bigints... Which haven't given me any problems. Wouldn't use it for client generated IDs but that is not what most applications require anyway.