This is a really excellent paper about the gaps between the way application developers perform validation and the ways that database authors expect them to.
I am curious whether AR implements uniqueness - begin -> read -> write -> commit - with 4 roundtrips to the database, or sends them all as one command.
I’m also curious whether AR implements save() by sending all fields to the database in an UPDATE, or only the changed ones? The former seems like it would be a good thing to discuss in the paper
It dirty-tracks changed attributes and updates only those. This is one of the seven ways you end up with invalid data in your AR database that I will be presenting at KCDC in June.
This is a really excellent paper about the gaps between the way application developers perform validation and the ways that database authors expect them to.
I am curious whether AR implements uniqueness - begin -> read -> write -> commit - with 4 roundtrips to the database, or sends them all as one command.
I’m also curious whether AR implements save() by sending all fields to the database in an UPDATE, or only the changed ones? The former seems like it would be a good thing to discuss in the paper
It dirty-tracks changed attributes and updates only those. This is one of the seven ways you end up with invalid data in your AR database that I will be presenting at KCDC in June.