JSON Schema file input too complex

[scylladb]

JSON Schema file input too complex

[dahankzter]

Actually it's not that bad when I think about it. A UDT for example looks like this in JSON:

{
      "Name": "col0",
      "Type": {
              "Types": {
                            "udt_672245080_0": "ascii",
                            "udt_672245080_1": "boolean",
                            "udt_672245080_2": "bigint",
                            "udt_672245080_3": "blob"
                },
                "TypeName": "udt_672245080",
                "Frozen": true
          }
}

The UDT field names are ugly simply because they are generated but it is quite clear that it is a complex type. The fact that it is a UDT isn't explicit but a "map" of type names and simple types should be a clue. The generated names hint at it but when hand writing such a file they can of course be called anything.

A tuple looks like:

{
          "Name": "col5",
          "Type": {
                  "Types": [
                            "varchar",
                            "smallint"
                   ],
                   "Frozen": false
          }
}

This is also pretty clear.

What does not seem clear is the index definitions:

{
         "Name": "col0_idx",
         "Column": {
                        "Name": "col0",
                        "Type": {
                            "Types": {
                                "udt_672245080_0": "ascii",
                                "udt_672245080_1": "boolean",
                                "udt_672245080_2": "bigint",
                                "udt_672245080_3": "blob"
                            },
                            "TypeName": "udt_672245080",
                            "Frozen": true
                    }
         }
}

This refers to the column col0 and repeats the type information. This should be omitted and be a simple reference by either column name or column index.

[dahankzter]

The index definition can look like:

"indexes": [
                {
                    "Name": "col0_idx",
                    "Column": "col0"
                },
                {
                    "Name": "col1_idx",
                    "Column": "col1"
                },
                {
                    "Name": "col2_idx",
                    "Column": "col2"
                },
                {
                    "Name": "col3_idx",
                    "Column": "col3"
                }
            ]

Without affecting functionality but greatly increasing readability.