When to use or not use the Contract Scope?

Question

Analyzing the eosio.token contract we see that the token stats is stored under the SYM scope. So if I want to read a token stat, like supply, I need to execute:

cleos get table eosio.token EOS stats

Replace EOS for any token that you create. My question is why it's architected in this way? Wouldn't be better if it was stored in the _self scope? Because if I run this instead:

cleos get table eosio.token eosio.token stats

I can have and work with a list of all created tokens under this contract and have a better vision. Actually how could I list and know all the created tokens if I don't know each SYM scope?

Answer 1

By using the scope you can have easier fetches. Think of it like the exact index you are searching for without have to use upper/lower bounds.

If we take that same logic to a user's info we could do

cleos get table contract username account cleos get table contract username balance cleos get table contract username friends

instead of cleos get table contract contract accounts -L lower -U upper

Logically, I'm also sure this is better resource wise as it only has to check for a single record, and not iterate the set.

Answer 2

Does anyone have the definition of scope ? To me, it basically looks like a way to compartmentalize data in a contract so that it only accessible within a certain space. In the above example, the data account balance and friends will exist in the contract only for the matching username given. When a different username is given as the scope, a new account balance and friends will exist for that usernames scope.

But also note that in that example account balance and friends are probably multi_index containers themselves and each can contain multiple data objects within them. So you could have multiple friends objects in your username scope.

In the token example, each account is used as the scope for the account balance. That way you have a container called accounts which holds multiple account objects, where each account object holds your balance for that given token symbol. This way when you query a users account using their scope you will get a list of all the tokens (every ticker symbol) that users has an account with a balance for.

I think they do it this way to keep all the token symbols that a users has in one place. Whereas if you scoped it to the contract itself, you would have to redesign the two containers accounts and currencys_stats where each account would need to hold each new token created, like with a vector or something. I guess this makes it easier to separate accounts from each other instead of getting a giant list of every account with every token symbol in existence. And easier to know who has to pay for the storage of each new token symbol added. How did they knew to design it like this in first place? Idk

Answer 3

The scope can be used to narrow down a table list to contain only some fraction of the total entries available.

However, you can also have the best of both worlds if you are willing to pay for the RAM involved.

Why not create a scope called global, which contains all of the entries from each individual scope, or alternatively contains a list of each scope name. Then you have a way to loop through several scopes at once as appropriate.