006: Drop support for string IDs
Supporting string IDs has made the source code more complex because it:
- We have to handle each case
- The signature of the functions is more complex.
strids don't have the auto-increment feature.
On the other hand, string IDs are useful when the identifying property of an entity is a string. This gives two benefits:
- You can
repo.getby that element.
- You can't have two elements with the same attribute.
We can try to simplify the ID signature to
int only while giving the users
tools so that they don't loose the benefits of the
- We already added an optional argument to
repo.getso that you can get by another attribute. This way it would be easy to spot what attributes would need their own index.
- We'll need to support uniqueness of entities based on an attribute other than
id_. (We need to refine this idea.)