Avoid table replacement with the changing of schema name

[andreqaugusto]

Minimal Example Code:

resource "snowflake_database" "my_db" {
  name = "my_db"
}

resource "snowflake_schema" "my_schema" {
  database = "my_db"
  name     = "FOO"
  comment  = "My FOO schema"
}

resource "snowflake_table" "my_table" {
  database =  "my_db"
  schema   = snowflake_schema.my_schema.name
  name     = "PEOPLE"
  comment  = "The people table"
  column {
    name    = "NAME"
    type    = "VARCHAR"
  }
}

Corresponding resources:

Changing the snowflake_schema.my_schema name from FOO to something else (e.g. BAR) will destroy the current table and create a new one (a forced replacement).

And the table will lose all data:

This is obviously not ideal, especially because Snowflake allows you to change a Schema name without losing the table data. In other words, if I wasn't tracking the PEOPLE table with terraform, I wouldn't lose any data and the table would gracefully live in the BAR schema as desired.

Note that you can actually bypass this issue with the following steps (that is what I did when faced with such scenario):

1. use terraform state rm snowflake_table.my_table to remove the tracked my_table
2. change the my_schema name to what you desire (e.g. from FOO to BAR)
3. use terraform import snowflake_table.my_table MY_DB|BAR|PEOPLE to import the table with all of our data in the new schema.

Obviously the steps above are not ideal (since we are messing with the state, which is always not recommended). The solution I'd like involves doing the above behind the scenes. I don't know if that is possible, but I suspect to be so (since, like I said before, Snowflake allows you to change schema names without any table replacement).