title: Simple blogging engine
published: 2025-02-18previous [‘Blog Questions Challenge 2025’] next [‘I quit everything’] parent directory
As mentioned in the previous post, I have been using several frameworks for blogging. But the threshold to overcome to begin and write new articles were always too hight to just get started. Additionally, I’m getting more and more annoyed by the internet, or specifically browsing the www via HTTP/S. It’s beginning to feel like hard work to not get tracked everywhere and to not support big tech and their fascist CEOs by using their services. That’s why I found the gemini protocol interesting ever since I got to know about it. I wrote about it before:
That’s why I decided to not go for HTTPS-first in my blog, but do gemini first. Although you’re probably reading this as the generated HTML or in your feed reader.
To just get started, I’m now using my tmux session that is running 24/7 on my home server. It’s the session I open by default on all my devices, because it contains my messaging (IRC, Signal, Matrix) and news (RSS feeds). Not it also contains a neovim session that let’s me just push all my thoughts into text files easily and everywhere.
The format I write in is gemtext, a markup language that is even
simpler as Markdown. Gemtext allows three different headings, links,
lists, blockquotes and formatted text, and that’s it. And to make my
life even easier, I only need to touch a file
.directory-listing-ok to let agate create an autoindex of
each directory, so I don’t have to take care about house-keeping and
linking my articles too much. I just went with this scheme to make sure
my posts appear in a correct order:
blog
└── 2025
├── 1
│ └── index.gmi
└── 2
└── index.gmi
When pointed to a directory, agate will automatically serve the
index.gmi if it finds one.
To serve the files in my gemlog, I just copy them as is, using
rsync. If anyone would browse the gemini space I would be
done at this point. I’m using agate, a gemini server
written in Rust, to serve the static blog. Technically, gemini would
allow more than that, using cgi to process requests and dynamically
return responses, but simple is just fine.
However, if I ever want any person to actually read this, sadly I
will have to offer more than gemtext. Translating everything into HTML
and compiling an atom.xml comes with some more challenges. Now I will
need some metadata like title and date. For
now I’m just going to add that as formatted text at the beginning of
each file I want to publish. The advantage is, that I can filter out
files I want to keep private this way. Using ripgrep I just find all
files with the published directive and pipe them through my
publishing script.
To generate the HTML, I’m going the route
gemtext -> markdown -> html, in lack of better ideas.
Gemtext to Markdown is trivial, I only need to format the links (using
sed in my case). To generate the HTML I use pandoc,
although it’s way too powerful and not-lightweight for this task. But I
just like pandoc. I’m adding simple.css to I don’t have to fuddle around
with any design questions.
I was looking for an atom feed generator, until I noticed how easily this file can be generated manually. Again, a little bit of ripgrep and bash leaves me with an atom.xml that I’m actually quite happy with.
I hope I have put everything out of the way get started easily and quickly. I could configure the system until the end of time to make unimportant things look better, but I don’t want to fall into that trap (again). I’m going to publish my scripts to a public repository soon, in case anyone feels inspired to go a similar route.