User:Stevestreza.com

From IndieWeb

Steve Streza

(photo)

Software developer, comic book writer, video creator, hobbyist located in Seattle, WA.

https://stevestreza.com

Elsewhere: Twitter GitHub YouTube

My name is Steve Streza. I love learning and making things, whether itโ€™s software, hardware, computers, wood projects, electronics, or whatever else I get my hands on.

My website is built on gatsby and is a work in progress.

Go Sounders!

Itches

  • โœ… Setup microformats across my site that are automatically generated
  • Setup POSSE through a variety of platforms. Don't expect readers to come to me, go to the services they already use.
    • Telegram
    • Twitter
    • Apple News
    • Medium
    • LinkedIn
    • SMS
    • Email
  • Configure website to use a push-driven CDN (probably KeyCDN)
  • โœ… Setup WebSub for important resources
    • โœ… Later setup WebSub for literally every page
  • Setup a web-based editor to create posts (e.g. Netlify CMS)
  • Support Micropub for publishing posts

How It's Made

Static Site Generation

My website is built with gatsby and has the full website, all the post content, and all microservices in a single git repository. When a change is made to any of those pieces, the change is committed and pushed to my personal GitLab instance, where a CI job builds the entire site, including a JSON file that has the entire contents of every page and post on the site. Assuming this is successful, the website is deployed to the server, and a webhook is sent to a microservice running on AWS Lambda to kickoff syndication.

Syndication

The syndication webhook is stored on AWS Lambda, and is connected to AWS DynamoDB. A full copy of all the website data is stored in a database. When the syndication webhook is triggered, a Node.js function is called that downloads the JSON site data file and compares it to DynamoDB to see if the post was created, updated, deleted, or stayed the same, and then updates the database as appropriate by creating, updating, and deleting records.

DynamoDB supports streaming changes as an event log, and other Lambda functions can trigger when create/update/delete events are added to the stream. This means support for POSSE syndication can be added just by hooking into the stream, filtering to the types of events that are needed, and then calling the appropriate APIs to do the syndication. For example, a Telegram syndication would filter for create events of posts, transform those posts into API calls, and then perform those API calls, but it wouldn't bother with updating or deleting posts. But a Medium syndication should update and delete posts if I change them in the source location, so that syndicator would listen to all of the events separately and perform the changes as needed. This should scale to as many different services as I want to syndicate to, now or in the future, and crucially takes the burden of ops out of my hands.

Events