pyjulapi

A demonstration of how to create a Python API that also calls Julia code.

Quick start

1. Install docker
2. Run docker-compose up from this repo's top directory
3. Create a job with curl -X POST "localhost:8000/job", which will return {"job_uuid": "57ca5074-05cc-4a60-830a-9af9a1b8641e"}
4. Copy the job_uuid and pass it to the result endpoint with curl -X GET "localhost:8000/result?job_uuid=57ca5074-05cc-4a60-830a-9af9a1b8641e"

Summary

• The Python Django package is used to create the external API (exposes port 8000).
• Within Django we use Postgres to store the data.
• Within Django we use Celery to manage asynchronous tasks, specifically a python task called runjob is kicked off in the job view
• (Redis is used as the task queue for Celery)
• The runjob task calls an internal Julia API to further process the job.
• Results are updated in the Postgres database, which the user can receive via the result endpoint.