Validating a deeply nested schema using custom "strict" draft 2020 metaschema can be very slow

[cweiske]

I am using check-jsonschema 0.23.0 with jsonschema 4.17.3 to validate a JSON schema against a strict json schema schema.

When running check-jsonschema, it runs for at least a minute until I kill it. Looks like an infinite loop somewhere.
Validating against the "normal" json schema schema works.

$ check-jsonschema --schemafile ~/dev/tools/json-schema-schema/json-schema.schema.json api-rest-connect-stick-stick-xxx.schema.json 
ok -- validation done

$ check-jsonschema --schemafile ~/dev/tools/json-schema-schema/strict.json api-rest-connect-stick-stick-xxx.schema.json 
(ctrl-c pressed after 1 minute)

The schema I am validating against is this: https://github.com/orgs/json-schema-org/discussions/380#discussioncomment-5711007
The schema file I am writing (and which was generated by genson) is attached here: api-rest-connect-stick-stick-xxx.schema.json.gz

[Julian]

Hey! Did you mean to file this on the check-jsonschema repo? If not, can you minimize to something that doesn't go via the CLI?

[cweiske]

I thought that check-jsonschema uses jsonschema internally, and that this problem is most likely a bug in jsonschema.
But since you insist I'll close this bug and create on in check-jsonschema.

[Julian]

I didn't insist -- or tell you to move it, I said if you think it's a likely jsonschema bug, please minimize your example :)

[sirosen]

Can/should we reopen this? I don't really know what to do with the CLI ticket until/unless we determine that it is not a jsonschema issue, so I'm thinking just keep both open for a minute.

I've pulled down the schema and instance the user provided and validated that this does at least run for a long time for me -- 6 minutes -- but that it terminates and in fact succeeds.

Just using the CLI -- I don't have time to try harder to give a python-only repro this minute -- I was able to reproduce a slow validation pass by trimming the user's schema down to the following document:

{
  "type": "object",
  "properties": {
    "body": {
      "type": "object",
      "properties": {
        "config": {
          "type": "object",
          "properties": {
            "global": {
              "type": "object",
              "properties": {
                "uitranslation": {
                  "type": "object",
                  "properties": {
                    "country": {
                      "type": "array",
                      "items": {
                        "type": "object",
                        "properties": {
                          "language": {
                            "type": "array",
                            "items": {
                              "type": "object",
                              "properties": {
                                "entries": {
                                  "type": "array",
                                  "items": {
                                    "type": "object",
                                    "properties": {
                                      "text": {
                                        "type": "string"
                                      }
                                    }
                                  }
                                }
                              }
                            }
                          }
                        }
                      }
                    }
                  }
                }
              }
            }
          }
        }
      }
    }
  }
}

and the metaschema from that thread was this:

{
  "$id": "https://example.com/draft/2020-12/schema/strict",
  "$schema": "https://json-schema.org/draft/2020-12/schema",

  "$vocabulary": {
    "https://json-schema.org/draft/2020-12/vocab/core": true,
    "https://json-schema.org/draft/2020-12/vocab/applicator": true,
    "https://json-schema.org/draft/2020-12/vocab/unevaluated": true,
    "https://json-schema.org/draft/2020-12/vocab/validation": true,
    "https://json-schema.org/draft/2020-12/vocab/meta-data": true,
    "https://json-schema.org/draft/2020-12/vocab/format-annotation": true,
    "https://json-schema.org/draft/2020-12/vocab/content": true
  },
  "$dynamicAnchor": "meta",

  "$ref": "https://json-schema.org/draft/2020-12/schema",
  "unevaluatedProperties": false
}

For me, on jsonschema v4.17.3, validating this takes just over a minute.
I haven't profiled it at all, but I'm guessing that a good chunk of time might be spent on ref resolution for the draft 2020 schema if that's complex on the inside.

[Julian]

The example here seems to complete in around ~1 second on my machine on 4.18.3. I'm going to call whatever was happening here fixed via the performance optimizations that went in with the referencing implementation, but if you're still seeing an issue, speak up.