4 ms·

"The existence of the theory implies that the “composed” design somehow arises, naively(?) or naturally." Given how many first timers come up with really weird

"The existence of the theory implies that the “composed” design somehow arises, naively(?) or naturally." Given how many first timers come up with really weird schema designs, I'm not necessarily surprised, although I agree presenting it as perhaps a default-but-wrong approach doesn't help much.

anonzzzies 2y ago
Not only beginners. I work with large teams with almost only seniors (10+ years of programming job experience) for large corps who don't know what 4nf is; they just chuck stuff in a table, update the same data across multiple tables, don't use indices or cascading rules etc. Everything is in code; the db just stores it, badly. They threat everything like it's mongo and this doesn't even work in mongo (no indices and it dies as well of course). 4NF was explained confusingly because it's quite complex theory. The fact the author says 'to be best of his understanding' it is 4nf is kind of telling. With many things in software dev; it all doesn't matter until it matter. With few records and few users, like the average department LoB crud thing, it's not really a problem; it will be when either or both grow. My 'colleagues' (I am external) claim to know all this stuff, but in reality there is no proof they do; I have never seen them do things correctly with respect to dbs or software in general. That makes me a lot of money, but it is depressing really.
- petalmind 2y ago
  > 4NF was explained confusingly because it's quite complex theory. The fact the author says 'to be best of his understanding' it is 4nf is kind of telling. > My 'colleagues' (I am external) claim to know all this stuff, but in reality there is no proof they do; I have never seen them do things correctly with respect to dbs or software in general. Wait I don't get it. Are you saying that "(instructor_id, skill_id); PK (instructor_id, skill_id)" (and the same for languages) is NOT in 4NF; and I am not right? What is in 4NF, then?
  - anonzzzies 2y ago
    I think you misunderstood me? I am talking about my colleagues... Or the many different colleagues I have throughout the year.
    - petalmind 2y ago
      Ah, okay, yes, sorry, thank you!
- mgkimsal 2y ago
  I was initially going to add some stuff I find a bit... shocking in the 2020s related to what I think are basic database stuff (maybe not specifically 4nf related). Held off, but I'll add one. A colleague had to argue that an 'address' table should have a synthetic primary key, instead of a natural key. The argument for the initial 'design' held that 'the address itself is unique - there's not 2 '123 west main streets'!". Wasn't my company directly, but the crux of the conversation was shared with me, and ... in 2022, with however many articles, tutorials, books, classes, stackoverflow, etc... having to argue that 'no, someone's entire address should not be the natural key for a table' was ... flummoxing to me.

anonzzzies 2y ago

Not only beginners. I work with large teams with almost only seniors (10+ years of programming job experience) for large corps who don't know what 4nf is; they just chuck stuff in a table, update the same data across multiple tables, don't use indices or cascading rules etc. Everything is in code; the db just stores it, badly. They threat everything like it's mongo and this doesn't even work in mongo (no indices and it dies as well of course). 4NF was explained confusingly because it's quite complex theory. The fact the author says 'to be best of his understanding' it is 4nf is kind of telling. With many things in software dev; it all doesn't matter until it matter. With few records and few users, like the average department LoB crud thing, it's not really a problem; it will be when either or both grow. My 'colleagues' (I am external) claim to know all this stuff, but in reality there is no proof they do; I have never seen them do things correctly with respect to dbs or software in general. That makes me a lot of money, but it is depressing really.

petalmind 2y ago
> 4NF was explained confusingly because it's quite complex theory. The fact the author says 'to be best of his understanding' it is 4nf is kind of telling. > My 'colleagues' (I am external) claim to know all this stuff, but in reality there is no proof they do; I have never seen them do things correctly with respect to dbs or software in general. Wait I don't get it. Are you saying that "(instructor_id, skill_id); PK (instructor_id, skill_id)" (and the same for languages) is NOT in 4NF; and I am not right? What is in 4NF, then?
- anonzzzies 2y ago
  I think you misunderstood me? I am talking about my colleagues... Or the many different colleagues I have throughout the year.
  - petalmind 2y ago
    Ah, okay, yes, sorry, thank you!
mgkimsal 2y ago
I was initially going to add some stuff I find a bit... shocking in the 2020s related to what I think are basic database stuff (maybe not specifically 4nf related). Held off, but I'll add one. A colleague had to argue that an 'address' table should have a synthetic primary key, instead of a natural key. The argument for the initial 'design' held that 'the address itself is unique - there's not 2 '123 west main streets'!". Wasn't my company directly, but the crux of the conversation was shared with me, and ... in 2022, with however many articles, tutorials, books, classes, stackoverflow, etc... having to argue that 'no, someone's entire address should not be the natural key for a table' was ... flummoxing to me.