Set iterator type for each index parts

[NERON]

Having index:
box.space.feeds:create_index('primary', { type = 'TREE',unique = true, parts = {1, 'unsigned',2,'unsigned',3,'unsigned'}})
I'm using pairs for iterating over index, but how can i set different iterator type for each parts: (EQ parts[1], LE parts[2],GE[3])?
In SQL we can set "order by" order for each field, but how we can do it by tarantool without writing special lua functions?

[kostja]

This is actually a very good idea, and we need it in SQL as well.

[Khatskevich]

Should this iterator walk only over a continuous range of tuples? Or skip unsuitable tuples?

E.g imagine data:

1. [1, 2]
2. [1, 3]
3. [2, 1]
4. [2, 2]
5. [2, 3]

What should the iterators return?:

1. [GE=1, EQ=2] ([[1,2]] or [[1,2][2,2]]?)
2. [GE=1, LE=1] ([[2,1]] or [[]]?)
3. [GE=1, LE=2] ([[1,2]] or [[1,2],[2,2]]?)
4. Some other (more complex) cases.

[locker]

Should this iterator walk only over a continuous range of tuples? Or skip unsuitable tuples?

E.g imagine data:

1. [1, 2]
2. [1, 3]
3. [2, 1]
4. [2, 2]
5. [2, 3]

What should the iterators return?:

1. [GE=1, EQ=2] ([[1,2]] or [[1,2][2,2]]?)

[[1,2][2,2]]

2. [GE=1, LE=1] ([[2,1]] or [[]]?)

[[2,1]]

3. [GE=1, LE=2] ([[1,2]] or [[1,2],[2,2]]?)

[[1,2],[2,1],[2,2]]

In other words, a multi-directional iterator should return all tuples matching the given criteria, even if they don't constitute a singe range. That would match the behavior of SQL SELECT statement.