[BUG] multiple schemas that contains tables by the same name raise decoding error.

[marxus]

[BUG] multiple schemas that contains tables by the same name raise decoding error.

[marxus]

nice, this is actually very elegant way to express it.

[marxus]

btw i'm not sure about HCL interpolation internals. but from working on terraform this is an option on how to avoid this matter

table "some_table_1" { /* name for reference */
  name = "some_table" /* actual table_name */
  schema = schema.some_schema_1
  column "key" { type = varchar(255) }
  column "value" { type = longtext }
  primary_key {
    columns = [table.some_table_1.column.key]
  }
  index "key" {
    unique  = true
    columns = [table.some_table_1.column.key]
  }
}
table "some_table_2" { /* name for reference */
  name = "some_table" /* actual table_name */
  schema = schema.some_schema_2
  column "my_key" { type = varchar(255) }
  column "value" { type = longtext }
  primary_key {
    columns = [table.some_table_2.column.my_key]
  }
  index "my_key" {
    unique  = true
    columns = [table.some_table_2.column.my_key]
  }
}
schema "some_schema_1" {}
schema "some_schema_2" {}

[rotemtam]

@marxus thanks for reporting this. I'm aware of this issue which results from the way we load symbols into the HCL context for parsing. We will work on this and update

[rotemtam]

hey @marxus,

This issue is solved by recent work we did on supporting qualifiers for tables. You can read about it in the doc change.

You can get the latest (canary) version of atlas using brew:

brew upgrade ariga/tap/atlas

Consider a situation where we have two schema a and b, and both of them contain a table named users, with schema qualifiers if you run:

atlas schema inspect -d "mysql://root:pass@localhost:3306/" --schema a --schema b

You get:

table "a" "users" {
  schema = schema.a
  column "id" {
    null = true
    type = int
  }
}
table "b" "users" {
  schema = schema.b
  column "id" {
    null = true
    type = int
  }
}
schema "a" {
  charset = "utf8mb4"
  collate = "utf8mb4_0900_ai_ci"
}
schema "b" {
  charset = "utf8mb4"
  collate = "utf8mb4_0900_ai_ci"
}

As you see the table names are qualified with a prefix table "a" "users" and table "b" "users".

HTH

Closing this, let me know if you need anything else.