oracle/graal · error · ExperimentParserTypeError

Failed to parse experiment {experimentId} in {resource}: exp

Error message

Failed to parse experiment {experimentId} in {resource}: expected {objectName} to be an instance of {expectedType} but got "{actualObject}"

What it means

The central type-narrowing check of ExperimentJSONParser: JSONLiteral.asInstanceOf(clazz) throws ExperimentParserTypeError when the decoded JSON value is not of the requested Java type. Every asInt/asString/asMap/asList/asLong/asBoolean call on log JSON funnels through it, so any schema mismatch in a proftool or optimization-log JSON surfaces here with the experiment id, symbolic file path, offending property name, expected type, and actual value.

Source

Thrown at compiler/src/org.graalvm.profdiff/src/org/graalvm/profdiff/parser/ExperimentJSONParser.java:72

         */
        private final Object object;

        /**
         * The name of the literal.
         */
        private final String objectName;

        private JSONLiteral(Object object, String objectName) {
            this.object = object;
            this.objectName = objectName;
        }

        @SuppressWarnings("unchecked")
        private <T> T asInstanceOf(Class<T> clazz) throws ExperimentParserTypeError {
            if (clazz.isInstance(object)) {
                return (T) object;
            }
            throw new ExperimentParserTypeError(experimentId, file.getSymbolicPath(), objectName, clazz, object);
        }

        private <T> T asNullableInstanceOf(Class<T> clazz) throws ExperimentParserTypeError {
            if (object == null) {
                return null;
            }
            return asInstanceOf(clazz);
        }

        public boolean isNull() {
            return object == null;
        }

        public String asString() throws ExperimentParserTypeError {
            return asInstanceOf(String.class);
        }

        public boolean asBoolean() throws ExperimentParserTypeError {

View on GitHub (pinned to a66e9ccd1d)

Solutions

Use a proftool output and optimization logs generated by the same GraalVM build as the profdiff you run.
Inspect the reported {resource}/{objectName} in the JSON and make its type match the schema (integers unquoted, maps as objects, arrays as arrays).
Regenerate the experiment data rather than patching JSON by hand.

Defensive patterns

Strategy: try-catch

Validate before calling

// Lightweight schema check before full parsing
Object totalPeriod = proftoolMap.get("total_period");
if (!(totalPeriod instanceof Number)) throw new IllegalStateException("total_period must be numeric");

Try / catch

try {
    new ExperimentJSONParser(experimentId, fileView).parse();
} catch (ExperimentParserTypeError e) {
    // message: experiment, resource, property, expected vs actual type; fix that JSON field
}

Prevention

Generate and parse logs with the same GraalVM version.
Do not edit JSON logs with tools that retype values (quoting numbers).

When it happens

Trigger: A proftool JSON where 'total_period' is a string, an optimization tree node whose 'name' is a number, a list property holding an object, etc. — any asX() accessor invoked on a JSON literal of a different type.

Common situations: Mixing proftool/optimization logs from a GraalVM version whose JSON schema differs from the profdiff version parsing them; hand-modified logs; third-party tools writing the log fields with different types.

Understand the failure class

Parsing and encoding errors: unexpected token, malformed input — why parsers reject input and how to find the real culprit.

Related errors

AI-assisted analysis of oracle/graal@a66e9ccd1d (2026-08-14). Data as JSON: /api/errors/5531a1cf7e9da813. Report an issue: GitHub.