MxMMz

This repository contains the code to generate my website. It contains:

• The content
• The styling
• The scripts to develop & generate the website

Nothing special here. Only Clojure files containing some functions. What might be a bit special is all the things that are not here: No configuration files, no build tools, no frameworks or libraries, no conventions ... no magic.

Motivation

• It's time for me to have my own little space on the internet to share my own thoughts and coding experiments.
• I have created some blogs before via wordpress, gatsby, multiple markdown templating engines, etc. but the overhead is often too big for me. When I want to write, I want to write, and I don't want to 'negotiate' with broken dependencies, unclear documentation, or "super-simple" config file structures (that somehow always grow into turing-complete DSLs, often without backwards compatibility).
• It looks like large parts of the web are becoming completely unusable: cookie popups, pay-if-you-want-to-read-more popups, subscription popups, banners and commercials begging for your attention, a lot of user tracking in the background ... I want NONE of that on this website.
• I want stable links ... the platform I used to blog on (1) broke my URLs and (2) changed the tool-suite completely within a year. Other platforms I've blogged on have the same unreliability. I want more stability and robustness.

Design decisions

• Use the least possible dependencies
• Use a stable language and ecosystem -> Clojure
• Stay close to standards so when my sparse dependencies break I still have options. Therefore use hiccup style html (https://github.com/weavejester/hiccup)
• Create (interactive) code examples (and their tests) in other repositories and only include the generated code in this repository, so this repository stays clean.
• Use stable links. Compound interest.
• Don't get too dependent on the host (currently: github pages). I've succeeded in also uploading the blog to my (domain-name) host while keeping all URLs the same and changing almost nothing except for the deployment target. So there is at least 1 backup strategy.

Setup

To work with this code you need these dependencies:

• Babashka: https://github.com/borkdude/babashka#installation
• Optionally: Babashka filewatcher pod: https://github.com/babashka/pod-babashka-filewatcher

Once those are installed and working, you can develop and generate the website.

To generate the website, call the gen.clj script from the rootfolder. It will generate the website in a publish folder, ready to be published.

To develop the website, call the dev.clj script from the rootfolder. It will start a watcher that will call the gen.clj script whenever a file is saved.

Publish

Short story: just push the generated html to the gh-pages branch

Longer story:

For publishing I'm basically using this setup: https://medium.com/linagora-engineering/deploying-your-js-app-to-github-pages-the-easy-way-or-not-1ef8c48424b7

It depends on a particular git configuration and because the web is a rough place these days (and websites may go down or page URLs might change) I'm going to repeat (and extend) the procedure here.

Some preconditions:

• Make sure you have github pages up and running, instructions here: https://pages.github.com
• Now the big leap (for me) was to understand that you don't have to publish everything you want to publish via that particular github pages repository!
  • So while you need a username.github.io repository, you can still publish pages from another repository.
  • If that "other-repository" is called other-repository, then the index.html in that other repository will be published on https://username.github.io/other-repository
  • ... that is ... if you configure it to be published in the settings of other-repository
  • This can be done in settings -> options -> github pages where you'll have to point to the branch where the index.html can be found that github pages has to publish.
• To make it a little bit more convenient, the setup described in the medium post
  • publishes from the gh-pages branch
  • makes a setup where you don't have to change branches to publish the website. Simply work on master or develop, at some point run your publish script and then cd to the publish directory (or in the post the dist directory) to automagically end up in the gh-pages branch where the only thing to do is add, commit and push your newly generated website
• WARNING: the setup assumes:
  • you start off with a repository where there is no publish directory yet
  • where publish/ is not yet added to the .gitignore file
  • where you run the commands in the root folder of an initialized git repository:

# Create an orphan branch named gh-pages
git checkout --orphan gh-pages

# Remove all files from staging
git rm -rf .

# Create an empty commit so that you will be able to push on the branch next
git commit --allow-empty -m "Init empty branch"

# Push the branch
git push origin gh-pages

# Come back to master
git checkout master

# Add publish to .gitignore
echo "publish/" >> .gitignore

git worktree add publish gh-pages

That's it. Now you run ./gen.clj and whenever you think you're ready to publish, cd publish and add, commit and git push origin gh-pages. Execute cd .. to get back on the master branch.

Credits

• https://github.com/borkdude helped me a lot, not only with his awesome https://github.com/borkdude/babashka tool, but also with this particular setup. Thanks Michiel!

License

• Use the code however you want at your own risk. It is not copyrighted.
• If you reuse parts of the code or styling, I would love to hear about it!
• The posts and website design are my work and subject to normal copyright law.