The use of _columns in update.

[ponytailer]

For example:

book.name = "name"
book.age = 1

# Should get dirty columns(name, age and other on_update columns), then update them, not all
await book.update()

OR

# Should update the column which were changed, too
await book.update(name="name", age=1)

But:

book.name = "name"
book.age = 1

# don't want to update the age,  who want to do it ?
# of course, the arg `_columns` can fix this problem.
await book.update(_columns=["name"])

``

[collerek]

Hi,

Yes, this is exactly the reason you have _columns in the update - if you do not want to update all columns.
You either set it by hard or accept some kind of user input - in both cases you know which columns are affected.

Keeping a state for each field would have too big performance penalty and can be unreliable
i.e. if you create a new instance from another one like:

instance = Model(**old_instance.dict())
instance.update() # what is dirty here? pk is set (and exists in db) so update is a valid query