Poor debugging experience with relayer integration tests

[Farhad-Shabani]

Context

Debugging the relayer’s integration tests is currently cumbersome and time-consuming, which significantly slows down the development process. Several factors contribute to this challenge (ranked in order of their perceived impact):

1. Insufficient tracing in Starknet relayer and Hermes SDK #261

2. Missing Debug implementations in Starknet relayer and Hermes SDK #262

3. Nix sometimes exhibits flakiness, particularly with ARM64 (regardless of MacOS vs Linux), introducing additional hurdles on such platforms. Examples:

   • [dev env]: Running tests locally on Mac OS #224
   • Flaky nix package builds during CI's integration tests #239
   • [e2e] slow CI runs #246

4. Overly complex test codebase, often covering multiple scenarios in a single run, making it difficult to isolate specific issues.

   • Issues: Develop reusable setups for relayer's integration tests #174

5. Limited compiler and rust-analyzer feedback in relayer #263

6. Challenges with implementation and reference searches in IDE #264

7. Lack of clear debugging guidance, even a simple one, including where to look for chain logs or other diagnostic information.

   • Connected to: Add README with project overview and run instructions #199

Highlight

While this issue might seem like a quality-of-life improvement, a technical debt, not directly tied to deliverables, addressing it would significantly impact the productivity, can reduce the time developers spend troubleshooting bugs—many of which may turn out to have simple fixes—allowing to focus more on delivering stuff.