oracle/graal · error · RuntimeError

Neither variable nor library specified.

Error message

Neither {daCapoClasspathEnvVarName} variable nor {daCapoLibraryName} library specified.

What it means

DaCapo benchmark suite environment validation failure. validateEnvironment() checks daCapoPath(); when neither the suite's DaCapo classpath environment variable (e.g. DACAPO_23_11_MR2_CHOPIN) nor the DaCapo mx library is available, the suite cannot locate the benchmark JARs and aborts.

Solutions

1. Set the DaCapo classpath environment variable for your suite version, e.g. export DACAPO_23_11_MR2_CHOPIN=/path/to/dacapo-23.11-MR2-chopin
2. Alternatively ensure the DaCapo library is declared and downloadable in the mx suite (mx benchmark resolves it via daCapoLibraryName()) so daCapoPath() returns the library location.
3. Verify with a quick check that the path/JAR actually exists before re-running the benchmark.

Example fix

# before
mx benchmark dacapo-chopin:fop   # env var unset, library absent

# after
export DACAPO_23_11_MR2_CHOPIN=/opt/benchmarks/dacapo-23.11-MR2-chopin
mx benchmark dacapo-chopin:fop

import os
name = suite.daCapoClasspathEnvVarName()
if not os.environ.get(name) and suite.daCapoLibraryName() is None:
    raise SystemExit(f"Set {name} or provide the {suite.daCapoLibraryName()} library before running DaCapo.")

When it happens

Trigger: Running a DaCapo benchmark (mx benchmark dacapo-chopin:...) without the DaCapo classpath env var set and without the DaCapo library present/resolved in the mx suite's library dependencies.

Common situations: Fresh machine without the DaCapo archive downloaded; CI jobs that do not export the env var; suite.py library entry for DaCapo not loaded because the wrong suite is on the classpath.

Related errors

• data location is not supported for suite version
• Suite runs only a single benchmark.
• Suite runs only a single benchmark, got
• The SPECJBB2015 environment variable was not specified.
• The SPECJVM2008 environment variable was not specified.

AI-assisted analysis of oracle/graal@a66e9ccd1d (2026-08-14). Data as JSON: /api/errors/df00dc140bdcc9e9. Report an issue: GitHub.

Appendix: source

Thrown at sdk/mx.sdk/mx_sdk_benchmark.py:2574

    def daCapoIterations(self):
        raise NotImplementedError()

    def daCapoSizes(self):
        raise NotImplementedError()

    def completeBenchmarkList(self, bmSuiteArgs):
        return sorted([bench for bench in self.daCapoIterations().keys() if self.workloadSize() in self.daCapoSizes().get(bench, [])])

    def existingSizes(self):
        return list(dict.fromkeys([s for bench, sizes in self.daCapoSizes().items() for s in sizes]))

    def workloadSize(self):
        raise NotImplementedError()

    def validateEnvironment(self):
        if not self.daCapoPath():
            raise RuntimeError(
                "Neither " + self.daCapoClasspathEnvVarName() + " variable nor " +
                self.daCapoLibraryName() + " library specified.")

    def validateReturnCode(self, retcode):
        return retcode == 0

    def postprocessRunArgs(self, benchname, runArgs):
        parser = argparse.ArgumentParser(add_help=False)
        parser.add_argument("-n", "--iterations", default=None)
        parser.add_argument("-sf", default=1, type=float, help="The total number of iterations is equivalent to the value selected by the '-n' flag scaled by this factor.")
        parser.add_argument("-s", "--size", default=None)
        args, remaining = parser.parse_known_args(runArgs)

        if args.size:
            if args.size not in self.existingSizes():
                mx.abort(f"Unknown workload size '{args.size}'. Existing benchmark sizes are: {','.join(self.existingSizes())}")

            if args.size != self.workloadSize():

View on GitHub (pinned to a66e9ccd1d)