Bug: Optional `command` Component Input Not Working—Fails When Argument Not Passed or Empty (AzureML SDK v2)

[gitgud5000]

• Package Name: azure-ai-ml
• Package Version: 1.27.1
• Operating System: Ubuntu 22.04.5 LTS
• Python Version: Python 3.12.11

Describe the bug
When defining a command component in AzureML SDK v2, a string input marked as optional=True with a default value (or not) does not behave as expected.
If I either omit the input or pass an empty string (""), I receive a Request is invalid and/or missing fields. The Value field is required. error, even though the documentation states this is supported.

To Reproduce
Steps to reproduce the behavior:

1. Define a component input as optional:

from azure.ai.ml.entities import Input

inputs = {
    "run_date" : Input(type=STRING),
    "tags" : Input(type=STRING, optional=True, default="yolo")
}

2. Use the following command in the component:

arguments = [
    "--run_date ${{inputs.run_date}}",
    "$[[--tags ${{inputs.tags}}]]"
]

3. Submit a job without specifying "tags" in inputs_run, or set "tags": "".

python_step_component.compute = COMPUTE_TARGET_RUN
inputs_run = {
    "run_date": "2023-01-31",
    }
# Submit the job
job = ml_client.jobs.create_or_update(
    python_step_component(
        **inputs_run
    ),
    experiment_name="test-optional-args",
    
)

4. Error occurs:

HttpResponseError: (UserError) Request is invalid and/or missing fields.  
(RequestInvalid) The Value field is required.

Click for full error

HttpResponseError                         Traceback (most recent call last)
Cell In[6], line 13
      5 inputs_run = {
      6     "run_date": "***redacted****",
      7     "use_case": "***redacted****",
      8     "kedro_pipeline_to_run": "***redacted****",
      9     # "tags": "aa"  # Optional, default is ''
     10 }
     12 # 4. Submit the component as a standalone job
---> 13 job = ml_client.jobs.create_or_update(
     14     python_step_component(**inputs_run),
     15     experiment_name="nptb_bpyn_mt"
     16 )
     18 print(f"Submitted job name: {job.name}")

File /anaconda/envs/nptb_bpyn/lib/python3.12/site-packages/azure/core/tracing/decorator.py:138, in distributed_trace.<locals>.decorator.<locals>.wrapper_use_tracer(*args, **kwargs)
    136             for key, value in span_attributes.items():
    137                 span.add_attribute(key, value)  # type: ignore
--> 138             return func(*args, **kwargs)
    139 else:
    140     # Native path
    141     config = {}

File /anaconda/envs/nptb_bpyn/lib/python3.12/site-packages/azure/ai/ml/_telemetry/activity.py:371, in monitor_with_telemetry_mixin.<locals>.monitor.<locals>.wrapper(*args, **kwargs)
    369 dimensions = {**parameter_dimensions, **(custom_dimensions or {})}
    370 with log_activity(logger, activity_name or f.__name__, activity_type, dimensions) as activityLogger:
--> 371     return_value = f(*args, **kwargs)
    372     if not parameter_dimensions:
    373         # collect from return if no dimensions from parameter
    374         activityLogger.activity_info.update(_collect_from_return_value(return_value))

File /anaconda/envs/nptb_bpyn/lib/python3.12/site-packages/azure/ai/ml/operations/_job_operations.py:713, in JobOperations.create_or_update(self, job, description, compute, tags, experiment_name, skip_validation, **kwargs)
    707 if (rest_job_resource.properties.job_type == RestJobType.PIPELINE) or (
    708     hasattr(rest_job_resource.properties, "identity")
    709     and (isinstance(rest_job_resource.properties.identity, UserIdentity))
    710 ):
    711     self._set_headers_with_user_aml_token(kwargs)
--> 713 result = self._create_or_update_with_different_version_api(rest_job_resource=rest_job_resource, **kwargs)
    715 if is_local_run(result):
    716     ws_base_url = self._all_operations.all_operations[
    717         AzureMLResourceType.WORKSPACE
    718     ]._operation._client._base_url

File /anaconda/envs/nptb_bpyn/lib/python3.12/site-packages/azure/ai/ml/operations/_job_operations.py:761, in JobOperations._create_or_update_with_different_version_api(self, rest_job_resource, **kwargs)
    758 if rest_job_resource.properties.job_type == RestJobType.COMMAND:
    759     service_client_operation = self.service_client_01_2025_preview.jobs
--> 761 result = service_client_operation.create_or_update(
    762     id=rest_job_resource.name,
    763     resource_group_name=self._operation_scope.resource_group_name,
    764     workspace_name=self._workspace_name,
    765     body=rest_job_resource,
    766     **kwargs,
    767 )
    769 return result

File /anaconda/envs/nptb_bpyn/lib/python3.12/site-packages/azure/core/tracing/decorator.py:138, in distributed_trace.<locals>.decorator.<locals>.wrapper_use_tracer(*args, **kwargs)
    136             for key, value in span_attributes.items():
    137                 span.add_attribute(key, value)  # type: ignore
--> 138             return func(*args, **kwargs)
    139 else:
    140     # Native path
    141     config = {}

File /anaconda/envs/nptb_bpyn/lib/python3.12/site-packages/azure/ai/ml/_restclient/v2025_01_01_preview/operations/_jobs_operations.py:632, in JobsOperations.create_or_update(self, resource_group_name, workspace_name, id, body, **kwargs)
    630     map_error(status_code=response.status_code, response=response, error_map=error_map)
    631     error = self._deserialize.failsafe_deserialize(_models.ErrorResponse, pipeline_response)
--> 632     raise HttpResponseError(response=response, model=error, error_format=ARMErrorFormat)
    634 if response.status_code == 200:
    635     deserialized = self._deserialize('JobBase', pipeline_response)

HttpResponseError: (UserError) Request is invalid and/or missing fields.
Code: UserError
Message: Request is invalid and/or missing fields.
Exception Details:	(RequestInvalid) The Value field is required.
	Code: RequestInvalid
	Message: The Value field is required.
Additional Information:Type: ComponentName
Info: {
    "value": "managementfrontend"
}Type: Correlation
Info: {
    "value": {
        "operation": "dfc7960391fcc82d77649e2257ea3d0c",
        "request": "a7bae414cded0d35"
    }
}Type: Environment
Info: {
    "value": "eastus2"
}Type: Location
Info: {
    "value": "eastus2"
}Type: Time
Info: {
    "value": "2025-06-30T19:57:29.6789526+00:00"
}Type: InnerError
Info: {
    "value": {
        "code": "RequestInvalid",
        "innerError": null
    }
}Type: MessageFormat
Info: {
    "value": "Request is invalid and/or missing fields."
}

5. If "tags": "a" (non-empty), the job works.

Expected behavior
I expect that omitting the "tags" input, or passing an empty string, would work without error, as "tags" is defined as optional with a default.

This matches the official documentation.

Screenshots
N/A (errors shown in stack trace).

Additional context
This issue blocks workflows that expect empty or missing values to use the default.

[github-actions]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @Azure/azure-ml-sdk @azureml-github.

[swathipil]

Hi @gitgud5000 - Thanks for opening an issue! We'll take a look at this asap.