diegiwg / mycro-db Goto Github PK

Reading the Readme I found that one of the future plans is to create the index system.

Thinking a little, I believe I came up with a good basis for the system:

API

import { MemoryStorage, MycroDatabase } from 'mycro-db';

const storage = new MemoryStorage();
const db = new MycroDatabase(inMemoryStorage);

const User = db.collection('users', {
    name: String(),
    age: Number()
});
User.index('name');

User.insert([{
        name: 'Alice',
        age: 28
    },
    {
        name: 'Bob',
        age: 30
    }
]);

OBJECT

{
  collections: {
    users: {
      id: 2,
      documents: [
          { id: 1, name: 'Alice', age: 28 },
          { id: 2, name: 'Bob', age: 30 }
      ],
      indexes: {
        // REF: [key]: (value, id)[]
        name: [
          ('Alice', 1),
          ('Bob', 2)
        ]
      }
    }
  }
}

In summary, in addition to CRUD commands, the Collections API can offer a command to create an INDEX under a KEY that you have in your documents.

As MycroDB is not concerned with maintaining something structured, the system can be flexible, and in documents inserted into the collection that do not have the key that is treated as an index, it simply ignores it.

In the same meantime, it would be necessary to rewrite part of the functions to deal with indices. Perhaps in addition to GET by id, it will be possible to have a GET by index. And change the QUERY to search in indexes?

Anyway, I believe that the approximation above is very good for the current MycroDB API.

From what I understand, it is not possible to choose the 'type' of data that will be used as an ID in a given collection. I would like this possibility to be added to the API.

Another thing, looking at the code that generates the collection, I saw that it 'automatically' / 'forcibly' attaches the Document ID to its record. Although I haven't messed up anything so far, I believe that doing this without the programmer explicitly stating that he wants to do so is shooting himself in the foot.

Anyway, I would like to have the choice of what type of data I will have as an ID, and for this automated ID system in the Document to be replaced by something less intrusive.