I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

Closes #18816 #17218 #14168

We just use the Pretty IR, wrap it in a newtype, derive a serde instance for it.

We intentionally do NOT use serde already implemented for the proto output of the to_proto phase (i.e. when to_stream_prost / to_batch_prost are called). That's because it locks the plan type to the stream_proto / batch_proto type.

EXPLAIN FORMAT JSON should only change the format, and not determine the underlying plan type (logical / stream / batch / to_proto).

Explain terse doesn't need to be supported. Using the json output, we can just do (see before/after below):

 jq 'def extract: {name: .name, children: [.children[] | extract]}; extract' plan.json

    {
      "name": "LogicalProject",
      "fields": {
        "exprs": [
          "approx_percentile($expr1)"
        ]
      },
      "children": [
        {
          "name": "LogicalAgg",
          "fields": {
            "aggs": [
              "approx_percentile($expr1)"
            ]
          },
          "children": [
            {
              "name": "LogicalProject",
              "fields": {
                "exprs": [
                  "t.v1::Float64 as $expr1"
                ]
              },
              "children": [
                {
                  "name": "LogicalScan",
                  "fields": {
                    "columns": [
                      "t.v1",
                      "t._row_id"
                    ],
                    "table": "t"
                  },
                  "children": []
                }
              ]
            }
          ]
        }
      ]
    }

{
  "name": "LogicalProject",
  "children": [
    {
      "name": "LogicalAgg",
      "children": [
        {
          "name": "LogicalProject",
          "children": [
            {
              "name": "LogicalScan",
              "children": []
            }
          ]
        }
      ]
    }
  ]
}

Checklist

• I have written necessary rustdoc comments
• I have added necessary unit tests and integration tests
• I have added test labels as necessary. See details.
• I have added fuzzing tests or opened an issue to track them. (Optional, recommended for new SQL features Sqlsmith: Sql feature generation #7934).
• My PR contains breaking changes. (If it deprecates some features, please create a tracking issue to remove them in the future).
• All checks passed in ./risedev check (or alias, ./risedev c)
• My PR changes performance-critical code. (Please run macro/micro-benchmarks and show the results.)

• My PR contains critical fixes that are necessary to be merged into the latest release. (Please check out the details)

Documentation

• My PR needs documentation updates. (Please use the Release note section below to summarize the impact on users)

Release note

Public Preview: Do note that the structure and format of EXPLAIN FORMAT JSON is considered unstable, and may change overtime.

After this PR, a new explain option is supported

FORMAT (TEXT | JSON)

For FORMAT JSON, it will output json for logical and physical plans, while it ignores trace, distsql options.
For FORMAT TEXT, the output is the same as the current output format of explain.

Here's an example:

create table t(v1 int);
explain(physical, format json) create materialized view m1 as select * from t;

  {
   "name": "StreamMaterialize",
   "fields": {
     "columns": [
       "v1",
       "t._row_id(hidden)"
     ],
     "pk_columns": [
       "t._row_id"
     ],
     "pk_conflict": "NoCheck",
     "stream_key": [
       "t._row_id"
     ]
   },
   "children": [
     {
       "name": "StreamTableScan",
       "fields": {
         "columns": [
           "v1",
           "_row_id"
         ],
         "table": "t"
       },
       "children": []
     }
   ]
 }