How to install

1. Ensure you are running Python 3.10
2. Optional: create a virtual environment (conda env, etc)
3. Install dependencies: pip install -r requirements.txt
4. Install dev dependencies: pip install -r requirements-dev.txt
5. Run tests: pytest
6. Update .env file in the project root with URL to Acme and Madrid instrument service endpoints for your system
7. Start up Fast API: python -m uvicorn app.main:app --reload --host 0.0.0.0 --port 7070
8. Navigate to http://localhost:7070/docs
9. Open POST /program section
10. Click Try it out button
11. Paste your request
12. Press Execute
13. Alternatively, you may try curl

curl -X 'POST' \
  'http://localhost:7070/program' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
    "id": "",
    "initial_value": 10,
    "control_instrument": "Acme",
    "operations": [
        {
            "type": "Sum",
            "value": 120
        },
        {
            "type": "Mul",
            "value": 3
        },
        {
            "type": "Div",
            "value": 2
        }
    ]
}'

Design decisions

• Used Abstract Base Classes (ABCs) in order to support extending system to other types of instruments
• Used a generic Rest API instrument service client as both existing clients follow the same format
  • If new formats are needed, then another implementation can be developed based on ABC
• Used a generic pulse generation method but still allowed for flexibility to provide custom translators via ABCs
• Exposed functionality via Fast API
  • This is just what people use these days including myself in my current job
• Used a mapping for being able to expand instruction sets
• Went minimal on error handling and validation as instructed
• Assumed that programs will always be simple sets of instructions
  • Reality is harsher though and this is obviously an oversimplification

Findings

• Values are integers (including negative)
  • Confirmed with requests to Acme and Madrid services
  • Confirmed by code inspection in Acme and Madrid services repositories
• Paths to load and run endpoints are not the same between Acme and Madrid
  • Confirmed by inspecting OpenAPI documentation for these services
• Madrid Service was not runnable on my system without changes
  • I had to optimize imports to fix it
  • Please see below