yes. This is part of what I have say about "make a GENERAL relational language and code the (full) app on it".
Is unnecessary complexity to have the "schema" split in several places (the DB catalog, the table, the view(s), the SQL fragment, the struct/classes, the mappings).
In my mind this is all you will need:
rel City of
id:Pk
name:Str CHECK NotEmpty
just ONCE.
If also, we can express the relational operators "raw":
plan1 = Scan(City, WHERE id= 1)
plan2 = FindKey(City, WHERE id= 1)
plan3 = City ?where id= 1 |> toPlan (turn into one of the above)
then is possible to also get zero cost abstractions.
Is unnecessary complexity to have the "schema" split in several places (the DB catalog, the table, the view(s), the SQL fragment, the struct/classes, the mappings).
In my mind this is all you will need:
just ONCE.If also, we can express the relational operators "raw":
then is possible to also get zero cost abstractions.BTW: Slowly I'm building this:
http://tablam.org
Stuck on trait objects this weeks, probably will stick with enums!
reply