top of page
  • Writer's pictureDmitry Amelchenko

Day 3 (Sunday) [KOA2 API implementation continued]

  • Running into a deployment issue. There are seem to be a way too many npm dependencies for the micro instance to handle. As a solution trying to commit them into git -- that's not very elegant but hopefully works. As an alternative option, may consider switching the prod env to yarn -- I've read somewhere that it's possible with Elastic Beanstack.

  • Adding npm's to git didn't help, it's now failing with some permissions error. Let's investigate the yarn path.

  • Found some links on the web with example yarn.config file -- still failed. Over all, this solution tends to be a lot more complicated than I like it to be.

  • Last resort -- bigger instance.

Side Note: It amazes me, how much effort we put into trying to save few dollars on hardware here and there, completely forgetting that the most valuable resource is our time. I wasted 3 hours, which easily converts into hundreds of dollars (depending on how much you value your time). And the solution was to spend $10 extra per month on bigger instance -- dah.

  • Finally, the API is deployed, the DNS/SSL is properly configured: 

WiSaw on the web
WiSaw on the web

  • Next, stubbing out the test methods for a) getting images feed based on distance, b) getting an image by id for detailed view, c) deleting an image by id.

  • Apparently, it's not that easy to findAll by distance from a particular point, will have to spend more time on this.

  • Finally, 7 tests pass, image create, retrieve the feed sorted by distance (exclude the full image data from results for performance reasons), retrieve a detailed image by id, delete image, everything deploys to Elastic BeanStalk. Good progress today. Considering that I have gazillions errands to run around the house as well and only able to spend half my available time on this project -- I'm very happy about the pace the project is moving at.

Side Note: The most difficult thing was to implement the very first method, but given that the number of methods keeps growing -- definitely a huge benefit to doing it the right way with automated testing. Everything is moving along super smoothly now, I can not imaging having to get the iOS (or Android) UI built without the backend methods reliably working first.


Recent Posts

See All


bottom of page