support opentelemetry python

[FeritBaturAltinn]

with this pull request opentelemetry python instrumentation measurement becomes possible

[DaGeRe]

Could you please address the mentioned issues?

[DaGeRe]

Put this in .gitignore, do not delete the config file (except there is a specific reason)

[DaGeRe]

This shouldn't be necessary for Otel Python

[DaGeRe]

Could you name this MOOBENCH_BIN_PY (like in config.rc of Kieker-python)?

[DaGeRe]

Please try whether this can be removed as well.

[DaGeRe]

This line can probably just be removed if you use startZipkin and stopBackgroundProcess.

[DaGeRe]

Could you move these into label.sh?

[DaGeRe]

Don't duplicate

[DaGeRe]

This is still present

[FeritBaturAltinn]

I have removed the duplicate.

[DaGeRe]

Don't duplicate this - it should be defined from benchmark.sh already

[FeritBaturAltinn]

I have removed the duplicate.

[DaGeRe]

Thanks for the update, I've made two comments. Could you check them, remove the two duplicates, and respond to the existing comment? (If its not possible to address the comment, just reply there and describe why).

[shinhyungyang]

@FeritBaturAltinn

Hi, thanks for the efforts! adding a new framework entry is not easy, because MooBench is an established framework that runs several tracing agents. Also, Kieker-python is still under developement. So I'm encouraged to see the work.

From a quick glance, I want to make some points.

1. Regarding tools/pybenchmark/benchmark.py, I agree with your changes.
2. Do you run it on Windows, or specifically under Windows/Cygwin environment? Maybe you can say e.g., Windows/Cygwin instead of just stating Windows.
3. For other tracing agents, we don't use export, whereas most of your variables use export. Is it necessary, and is it related to your Windows/Cygwin environment, perhaps?
4. We should be relying on the MooBench infrastructure as much as possible. For example, many of your variables in benchmark.sh is already defined in ${MAIN_DIR}/init.sh. I suggest to include this code in your benchmark.sh:

if [ -f "${MAIN_DIR}/init.sh" ] ; then
	source "${MAIN_DIR}/init.sh"
else
	echo "Missing library: ${MAIN_DIR}/init.sh"
	exit 1
fi

Then, you don't need:

# config.rc
export RESULTS_DIR="$BASE_DIR/results-OpenTelemetry-python"
export RAWFN="$RESULTS_DIR/raw"

# benchmark.sh
export NUM_OF_LOOPS=${NUM_OF_LOOPS:-10}
export TOTAL_NUM_OF_CALLS=${TOTAL_NUM_OF_CALLS:-2000000}
export RECURSION_DEPTH=${RECURSION_DEPTH:-10}
export METHOD_TIME=${METHOD_TIME:-0}
export SLEEP_TIME=${SLEEP_TIME:-15}

[FeritBaturAltinn]

@shinhyungyang

Hello, thank you for the suggestions. I have refactored benchmark.sh to source init.sh as requested. I also removed export statements from config.rc and functions.sh and added the Windows/Cygwin path detection logic. Finally, I verified the benchmark still runs successfully with accurate measurements after the changes.

[DaGeRe]

Thanks for the PR. There are some small things to improve (especially formatting with shfmt), but since it's functional, I'll merge for now.