3 ms·

A couple things: This is admittedly a bit pedantic, but in E.F. Codd's original paper, he defined "relation" as the relation between a tuple (table row) and an

A couple things: This is admittedly a bit pedantic, but in E.F. Codd's original paper, he defined "relation" as the relation between a tuple (table row) and an attribute (table column) in a single table - https://en.wikipedia.org/wiki/Relation_(database) https://en.wikipedia.org/wiki/Relation_(database). I'm not sure of the author's intent, but the user table example (user, country_id ) might imply the relationship between the user table and the country table. It's a common misconception about what "relational" means, but tbh I'm fine with that since it makes more sense to the average developer. If you ever need to join sets of data in code, don't use nested loops - O(n^2). Use a map / dictionary. It's one of the few things I picked up doing Leetcode problems that I've actually needed to apply irl.

zzleeper 3y ago
I think it depends on the data. If it's presorted by the join variable then rolling the loop is faster. Also, if the index is too big for memory, then it might be faster to loop.
- Anon4Now 3y ago
  Yeah. Many database tables are too large for an in memory hash join. My comment was a not very well fleshed out tangential remark on the value of practicing DS&A problems. I know a lot of devs hate Leetcode style interviews. I get it. It's not fun. But contrary to what some people say, I have run into a fair number of situations where the practice helped me implement more efficient solutions.
fuy 3y ago
I think the blog author uses it the same way Codd does, that's why he's talking about two columns from the same table.
abtinf 3y ago
I think understanding that “relation” means the relationships of attributes to a primary key is a crucial, foundational concept. Without it, you can’t properly understand a concept like normalization, why it arises, and when it applies. You can’t even fully grasp the scope of “data corruption” (most developers have an extremely poor understanding of how broad the notion of corruption is in databases).

zzleeper 3y ago

I think it depends on the data. If it's presorted by the join variable then rolling the loop is faster. Also, if the index is too big for memory, then it might be faster to loop.

Anon4Now 3y ago
Yeah. Many database tables are too large for an in memory hash join. My comment was a not very well fleshed out tangential remark on the value of practicing DS&A problems. I know a lot of devs hate Leetcode style interviews. I get it. It's not fun. But contrary to what some people say, I have run into a fair number of situations where the practice helped me implement more efficient solutions.

fuy 3y ago

I think the blog author uses it the same way Codd does, that's why he's talking about two columns from the same table.

abtinf 3y ago

I think understanding that “relation” means the relationships of attributes to a primary key is a crucial, foundational concept. Without it, you can’t properly understand a concept like normalization, why it arises, and when it applies. You can’t even fully grasp the scope of “data corruption” (most developers have an extremely poor understanding of how broad the notion of corruption is in databases).