Implementing tables in lua to access specific pieces for later use

Question

I am trying to make a table store 3 parts which will each be huge in length. The first is the name, second is EID, third is SID. I want to be able to get the information like this name[1] gives me the first name in the list of names, and like so for the other two. I'm running into problems with how to do this because it seems like everyone has their own way which are all very very different from one another. right now this is what I have.

info = {
       {name = "btest",  EID = "19867", SID = "664"},
       {name = "btest1", EID = "19867", SID = "664"},
       {name = "btest2", EID = "19867", SID = "664"},
       {name = "btest3", EID = "19867", SID = "664"},
     }

Theoretically speaking would i be able to just say info.name[1]? Or how else would I be able to arrange the table so I can access each part separately?

Answer 1

There are two main "ways" of storing the data:

Horizontal partitioning (Object-oriented)

Store each row of the data in a table. All tables must have the same fields.

Advantages: Each table contains related data, so it's easier passing it around (e.g, f(info[5])). Disadvantages: A table is to be created for each element, adding some overhead.

This looks exactly like your example:

info = {
    {name = "btest",  EID = "19867", SID = "664"},
    -- etc ...
}

print(info[2].names) -- access second name

Vertical partioning (Array-oriented)

Store each property in a table. All tables must have the same length.

Advantages: Less tables overall, and slightly more time and space efficient (Lua VM uses actual arrays). Disadvantages: Needs two objects to refer to a row: the table and the index. It's harder to insert/delete.

Your example would look like this:

info = {
    names = { "btest", "btest1", "btest2", "btest3", },
    EID   = { "19867", "19867",  "19867",  "19867",  },
    SID   = { "664",   "664",    "664",    "664",    },
}

print(info.names[2]) -- access second name

So which one should I choose?

Unless you are really need performance, you should go with horizontal partitioning. It's far more common working over full rows, and gives you more freedom in how you use your structures. If you decide to go full OO, having your data in horizontal form will be much easier.

---

Addendum

The names "horizontal" and "vertical" come from the table representation of a relational database.

  | names | EID | SID |         | names |
--+-------+-----+-----+         +-------+       
1 |       |     |     |         |       |       --+-------+-----+-----+
2 |       |     |     |         |       |       2 |       |     |     |
3 |       |     |     |         |       |       --+-------+-----+-----+