This one is open for debate and would be a big change.
There are lots of areas where it would be helpful to separate out the key schemas of a table from the rest of the attributes.
Once an item is saved in the database, its key shouldn't change - If you call put with a different key, it will create a new item, which is probably not what you want a lot of the time. You can't call update with key properties - you need to specify the key in one parameter and the values to change in another. I think we should reflect this logic separation.
A big reason for doing this would be to better support single-table design. If you do this properly, the key is kind of distinct from the actual data anyway - see this video for more info on what I mean here
This one is open for debate and would be a big change.
There are lots of areas where it would be helpful to separate out the key schemas of a table from the rest of the attributes.
Once an item is saved in the database, its key shouldn't change - If you call
putwith a different key, it will create a new item, which is probably not what you want a lot of the time. You can't callupdatewith key properties - you need to specify the key in one parameter and the values to change in another. I think we should reflect this logic separation.A big reason for doing this would be to better support single-table design. If you do this properly, the key is kind of distinct from the actual data anyway - see this video for more info on what I mean here