A brief history of the UUID
June 9, 2017 2:03 PM Subscribe
"Ever since two or more machines found themselves exchanging information on a network, they’ve needed a way to uniquely identify things."
Awesome article, however, a nitpick appears:
"With it brought the first unique identifier in a network: the telephone number."
I'd guess that earlier analog communications networks had unique identifiers. For example, the chaski network of the Tawantinsuyu probably had some unique way of referring to each tambo.
posted by signal at 3:04 PM on June 9, 2017 [3 favorites]
"With it brought the first unique identifier in a network: the telephone number."
I'd guess that earlier analog communications networks had unique identifiers. For example, the chaski network of the Tawantinsuyu probably had some unique way of referring to each tambo.
posted by signal at 3:04 PM on June 9, 2017 [3 favorites]
I like that clock skew isn't enough to screw up uniqueness here, just ordering. Once your systems get big enough, something is going to have a bad clock, always.
Time is the worst dimension.
posted by The Gaffer at 3:52 PM on June 9, 2017 [2 favorites]
Time is the worst dimension.
posted by The Gaffer at 3:52 PM on June 9, 2017 [2 favorites]
Time is the worst dimension.It's not time itself that drives us programmers crazy, it's all the stuff that happens in it.
posted by Horkus at 4:04 PM on June 9, 2017 [4 favorites]
The worst bugs are always time bugs.
posted by Amplify at 4:29 PM on June 9, 2017 [2 favorites]
posted by Amplify at 4:29 PM on June 9, 2017 [2 favorites]
I like this guy's video for a decent introduction to the problem of time and timezones for non-coders.
posted by jenkinsEar at 4:29 PM on June 9, 2017 [3 favorites]
posted by jenkinsEar at 4:29 PM on June 9, 2017 [3 favorites]
KSUID would be really helpful for a system I'm currently working on.
Odds are you can write a function that concatenates a 32-bit timestamp with a 128-bit random number in less time than it'll take you to make it through that article.
posted by effbot at 4:30 PM on June 9, 2017 [4 favorites]
Odds are you can write a function that concatenates a 32-bit timestamp with a 128-bit random number in less time than it'll take you to make it through that article.
posted by effbot at 4:30 PM on June 9, 2017 [4 favorites]
Nice to see a mention of DCE, a technology that's been totally forgotten now.
posted by octothorpe at 5:05 PM on June 9, 2017
posted by octothorpe at 5:05 PM on June 9, 2017
Odds are you can write a function that concatenates a 32-bit timestamp with a 128-bit random number in less time than it'll take you to make it through that article.
Real Programmers™ make their own mistakes instead of learning from other people's!
posted by shponglespore at 5:35 PM on June 9, 2017 [10 favorites]
Real Programmers™ make their own mistakes instead of learning from other people's!
posted by shponglespore at 5:35 PM on June 9, 2017 [10 favorites]
6d08883e-dbe0-4ec6-8f9c-05829b5db03b
posted by 1970s Antihero at 6:07 PM on June 9, 2017
posted by 1970s Antihero at 6:07 PM on June 9, 2017
hey!
posted by 922257033c4a0f3cecdbd819a46d626999d1af4a at 6:21 PM on June 9, 2017 [16 favorites]
posted by 922257033c4a0f3cecdbd819a46d626999d1af4a at 6:21 PM on June 9, 2017 [16 favorites]
I feel obligated to link to blue_beetle's comment about MeFi's RSS feed glitch
posted by Nonsteroidal Anti-Inflammatory Drug at 7:40 PM on June 9, 2017 [1 favorite]
posted by Nonsteroidal Anti-Inflammatory Drug at 7:40 PM on June 9, 2017 [1 favorite]
92225703-3c4a-0f3c-ecdb-d819a46d626999d1af4a is no valid uuid, although some programs will incorrectly decode it as such.
Oh hell, that's probably a security hole in something or other. I'm going to have to report a bug aren't I?
posted by joeyh at 7:48 PM on June 9, 2017 [1 favorite]
Oh hell, that's probably a security hole in something or other. I'm going to have to report a bug aren't I?
posted by joeyh at 7:48 PM on June 9, 2017 [1 favorite]
It's not clear to me why, if you're going to all the bother to expand on UUIDs and make something new, incompatible, and larger, you'd be content to restrict yourself to a 100-year lifespan with a 32-bit time field. Is there some magic limit that you're sneaking under at 160 bits?
posted by Western Infidels at 7:55 PM on June 9, 2017 [2 favorites]
posted by Western Infidels at 7:55 PM on June 9, 2017 [2 favorites]
Probably trying to sneak into the norms of a generation of people trained on SHA1
posted by ead at 9:00 PM on June 9, 2017
posted by ead at 9:00 PM on June 9, 2017
Real Programmers™ make their own mistakes instead of learning from other people's!
Not sure I follow here; you're saying that if someone else is known to make mistakes in trivial code, the right solution is to use their code in your program, to make sure you inherit all their mistakes? I can see a few problems with that approach...
posted by effbot at 2:29 PM on June 11, 2017
Not sure I follow here; you're saying that if someone else is known to make mistakes in trivial code, the right solution is to use their code in your program, to make sure you inherit all their mistakes? I can see a few problems with that approach...
posted by effbot at 2:29 PM on June 11, 2017
« Older One-Liners With Tourette's | Just in case you thought that the New 52 was a bad... Newer »
This thread has been archived and is closed to new comments
I like that clock skew isn't enough to screw up uniqueness here, just ordering. Once your systems get big enough, something is going to have a bad clock, always.
posted by JoeBlubaugh at 2:44 PM on June 9, 2017