Workflow Try & Retry

[joaodaher]

Adds try/catch blocks and retry policies to enhance workflow resilience. This change introduces:

• Try/catch blocks for error handling, enabling workflows to gracefully manage exceptions and execute fallback logic.
• Retry policies for workflow steps, allowing automatic retries with configurable backoff strategies.

This enhancement improves the robustness and reliability of workflows by providing mechanisms to recover from transient failures and handle errors in a controlled manner.

---

Note

Introduce try/catch error handling and retry policies with full YAML/codegen support, new example workflows, and comprehensive tests/docs; bump version to 0.0.2.

• Core:
  • Add TryCatchStep, TryCatchBuilder, and try_catch API (core/error_handling.py); export via core/__init__.py and package __init__.py.
• Codegen:
  • Emit retry config (predicate/backoff) and timeout per step; generate try/except blocks for TryCatchStep.
  • Refactor workflow emission to process nested steps; preserve required headers and run-id capture.
• Examples:
  • Add examples/app/flows/resilient_payment.py and data_pipeline.py; update examples/app/main.py imports.
• Tests:
  • Add unit tests for Arg, RetryPolicy, step core/execution, workflow core, and try/catch.
  • Add codegen tests + fixtures for retry and try/catch (tests/codegen/...); update smoke runner script references.
• Docs:
  • Update README.md to mark Retries/Try-catch as supported and add roadmap; tweak CONTRIBUTING.md and smoke instructions; add .cursor/rules/writing-tests.mdc.
• Version:
  • Bump package to 0.0.2.

^{Written by Cursor Bugbot for commit df49371. This will update automatically on new commits. Configure here.}

[qltysh]

All good ✅

[Copilot]

Pull Request Overview

This PR introduces try/catch blocks and retry policies to enhance workflow resilience. The implementation adds comprehensive error handling mechanisms that allow workflows to gracefully handle exceptions and recover from transient failures through configurable retry strategies.

• Adds TryCatchStep and TryCatchBuilder for structured exception handling with fallback logic
• Implements RetryPolicy with configurable backoff strategies for automatic retries on failures
• Provides YAML code generation support for both try/catch blocks and retry configurations

Reviewed Changes

Copilot reviewed 24 out of 24 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
src/fastapi_cloudflow/core/error_handling.py	New module defining TryCatchStep and TryCatchBuilder classes for workflow exception handling
src/fastapi_cloudflow/codegen/workflows.py	Updates YAML generation to support try/catch blocks and retry policy emission
src/fastapi_cloudflow/init.py	Exports new try/catch functionality and helper functions
tests/unit/test_try_catch.py	Comprehensive unit tests for try/catch functionality
tests/unit/test_retry_policy.py	Unit tests for retry policy features
tests/codegen/test_try_catch_codegen.py	Tests for try/catch YAML code generation
tests/codegen/test_retry_codegen.py	Tests for retry policy YAML code generation
examples/app/flows/resilient_payment.py	Example workflow demonstrating retry and try/catch usage
examples/app/flows/data_pipeline.py	Complex data pipeline example with nested try/catch blocks

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Copilot]

The reraise_error step is appended to the except_steps list after it has already been assigned to try_catch_def['except']['steps']. This creates a mutation after assignment issue. The reraise step should be appended before the assignment on line 163, or the assignment on line 167 should be removed since it's redundant.

Suggested change

try_catch_def["except"]["steps"] = except_steps

[Copilot]

[nitpick] The comment mentions 'to avoid syntax issues' but this approach is overly complex and fragile. Consider using a proper async function definition or pytest.mark.asyncio instead of manually creating AsyncMock objects.

[Copilot]

[nitpick] The inline comments on lines 115, 117, and 119 appear to be implementation notes rather than helpful code documentation. These should be removed or converted to proper documentation that explains the business logic.

Suggested change

	output_model=PaymentResult, # Changed to match the actual output
	try_steps=try_primary.nodes,
	except_steps=[fallback_payment, log_failure], # Use actual step instances
	error_var="payment_error",
	raise_on_error=False, # Don't re-raise, we handled it
	output_model=PaymentResult,
	try_steps=try_primary.nodes,
	except_steps=[fallback_payment, log_failure],
	error_var="payment_error",
	raise_on_error=False,

[Copilot]

[nitpick] Using a hardcoded mock endpoint URL in production code is not ideal. Consider using environment variables or configuration for external service URLs, even in examples.

[Copilot]

[nitpick] The line is split unnecessarily with parentheses. This could be written as a single line or use proper line continuation for better readability.

Suggested change

	tc_step = (
	try_catch("custom-error-var").try_block(try_block).except_block(except_block, error_var="my_error").build()
	)
	tc_step = try_catch("custom-error-var").try_block(try_block).except_block(except_block, error_var="my_error").build()