Stream output from check_call to HTTP response

[eddieantonio]

Run the commands and see what fails from the comfort of your GitHub action!

See: https://flask.palletsprojects.com/en/1.1.x/patterns/streaming/

[andrewdotn]

If we get tired of figuring out how to bump timeouts for super-long-running HTTP requests, some options include:

• POSTing a deploy request returns a token. Then, periodically send GET requests with that token to the deploy server to get the current status plus logs of the deployment, until it completes.
  • Implementation possibility: The deploy server keeps logs in memory if there’s only one worker process, otherwise in a little sqlite database. Status responses contain a cursor / timestamp / position to be included on the next request, so that poll only return new log lines.
• Stream over a websocket to some command-line websocket client