{ "Manuel": "Serra" }
_# blog

Hello world!!

This is my very first blog post in my very first personal website. Today I want to share with you why I chose to have a personal space plus the philosophy and stack behind it. Hope you'll like it!

7 min read11/1/2018


As a software engineer, I never stop learning new things.

That’s literally what I love the most @work.

On the other side I also love to share this technical knowledge with people that can find it useful.

In that moment I really feel my purpose.

Also, I do want a place for my findings to be persisted. Because I tend to basically forget the thing.

Last but not least: at the start of my career the path was actually pretty canonical. I learned about J2EE, JDBC and Java Server Pages or JSP (ndr. and many many more). The usual stack, isn’t it? That was entertaining at first but quickly became a little boring. Luckily, I’ve had the opportunity to deep dive into other technologies like containers, microservices and a little bit of frontend development.

During this roller caster it grows on me that:

The lesser the code I wrote, the better.

And I hope that every software engineer on the planet will agree.

And that’s the point about this blog: having aq personal blog-like site without having infrastructure nor a database to maintain.

The stack

SSG or static site generators

To actually forget having a database or infrastructure you need to strip out all the unneeded things and aim for the MVP: aka to output the good and old basic html/css.

That means you need to consume and produce all the permutation of your site ahead-of-time (or ahead of navigation). That is: if we have 10k blog posts (btw you’re a proficient writer) we’ll need to traverse all the posts and save the resulting processed html and style during build-time, before going live.

That’s actually what static site generators are supposed to do: generate all your site pages with the least amount a work needed client side. All in all is 2018 and SPAs are pretty common. They are great and awesome but nonetheless can be pretty heavy weight on the browser because we need to download and parse all our bundle (and all the frameworks and dependencies we’re relaying on) before display the first pixel of your app, this can take some time if the bundle is not trivial or not optimized.

Disclaimer: you can optimize SPAs too obv (chunks, defer render, Suspense, lazy), but that’s not the point here.

Processing all the data aot can appear cumbersome, but SSG are pretty fast doing that.

And so my SSG of choice is Gatsby because:

  • I enjoy writing React components and the workflow in general (I really like the strict separation between frontend and backend, React is the V of the ‘old’ MVC)
  • Is actually pretty fast
  • With its plugin foundation you can basically plug’n’play cool tools and having a ton of goodies in a matter of minutes (some PWA goodies and service worker things are tricky!)

…but do you need GraphQL, seems over-engineered to me!

That was a real concern from one colleagues of mine, specifically from @pscanf (hint: his site also is statically generated from React) - since I’ve a great esteem of him and I totally agreed with him, that needs to be addressed.

But first, for those that are not familiar with Gatsby this is more or less its canonical folder structure:

├── README.md
├── contents
│   └── our db, kind of?
├── gatsby-config.js
├── gatsby-node.js
├── package.json
├── src
│   ├── React stuff...
│   └── templates
└── static
    └── assets...

That looks like a normal nodejs/React project apart from those 2 guys: gatsby-node.js and gatsby-config.js. The former is actually where the magic aoc happens - where we create our site pages statically - and where the misconception that you must use GraphQL begin.

Thats better explained while looking at the code:

const fs = require("fs");
const fm = require("front-matter"); // Markdown front-matter
const md = require("markdown-it")({ html: true }); // Markdown html renderer

// `createPages` is the method executed by Gatsby at built-time
// See overview:
//     https://www.gatsbyjs.org/docs/gatsby-lifecycle-apis/#bootstrap-sequence
exports.createPages = async ({ actions: { createPage } }) => {
  // Looks for dir containing all our Markdown posts
  const dir = "our directory/aka db";
  const postsFiles = fs.readdirSync(dir);

  postsFiles.forEach(post => {
    // Read text from file
    const text = fs.readFileSync(`${dir}/${post}`, "utf-8");

    // Extract the Markdown front-matter (attributes) and content (body)
    const { attributes, body } = fm(text);

    // Render Markdown content/body to html
    const html = md.render(body);

    // Create page! Yuppie!
    // path:         will be the URL of our post
    // component:    our React component
    // context:      object pass through for our React component
      path: `/${attributes.slug}`,
      component: require.resolve("./src/templates/post.js"),
      context: { attributes, content: html }

That way, our React component will access the data we fed into this way:

const Template = ({ pageContext: { attributes, content } }) => (
      {Object.keys(attributes).map(key => (
        <Attribute>{`${key}: ${attributes[key]}`}</Attribute>
    <Content dangerouslySetInnerHTML={content} />

Gently wrapped into the pageContext property. No GraphQL to care about!

So our beloved React can render and outputs the html/css we need to deliver on our CDN. Ready to serve for clients browsers.


  • Your site will be fast to render and powerful to code and evolve
  • PWA and service workers near out-of-the-box
  • No database or infrastructure to care about (only static file serving, the cheapest part of the stack)
  • Ant that’s it!

Hope you like this post, please support me on Steemit!

I’ve some ideas to talk you about, maybe microservices next?