5.1 KiB
| title |
|---|
| Brainshit |
Brainshit is my personal blog, with some occasional posts from other people I have given access to. My best friend used to write with me in the early years, but is now just an avid reader. Lord Vlad joined me there after we got to know each other through habitica.
Brainshit is now part of the cybrecluster.
History
-
2009-2010: A very blue and orange website hosted on Freeheberg, using code from a PHP tutorial on the Site du Zéro (OpenClassrooms's old name), and with .tk domain name.
-
2010-2012: Renamed it from my old nickname to BubbleLand, redesigned it by filling it with
<img>tags and scanning drawings from a notebook, and moved to Kelio.org, a French non-profit. -
2013-2015: Made the initial design using Amaya, then integrated it again into the old PHP tutorial code. Moved to Hostinger, another free PHP hosting thing. Rewrote the whole admin page using jQuery UI once I had learned it during an internship and even tried to make a JSON API.
-
2015-now: Rewrote from scratch using things I learnt from doing a PHP project at school and reusing that project in an internship. Articles can have just a summary, or a summary and content, or a summary and chapters. A simple permissions system is available, as well as comment moderation, private messaging between writers and an idea board.
Next rewrite
As per my cybrecluster rules, I want to move Brainshit away from PHP and MariaDB and use a static site. I wanted to do that long before I even thought about the cybrecluster. I had written a Python exporter that took in a MariaDB database export and generated Markdown files and some tar.gz archives for the features that I would drop during the transition.
I had considered at one point rewriting Brainshit in Python with Django, before coming up with the cybrecluster and deciding that Brainshit is simply not large enough to require something non-static. While I definitely am able to pull off that rewrite, I don't want to host it.
I am struggling to find the static site generator that will allow me to make my transition. I was considering moving Brainshit further away from a blog and more into a wiki, but now that I have this wiki, this feels less relevant; these change of plans definitely slowed me down.
I have considered using Zola. It looked like Hugo, but nicer to use, and I now am the maintainer of its Alpine package. But after using it at work, I realized its Markdown parser is not CommonMark-compliant, and has many issues that are making it as uncomfortable to use as my current parser, PHP Markdown Extra.
Current planned design
I have been using pandoc for this website as well as on breadpunk, and maybe
someday I'll also use it on tilde.town, so why not use it on Brainshit as well?
That is a bit more complex to do because pandoc only handles a single document
at a time. Existing pandoc-based static site generators feel too blog-oriented
and not extensible enough to be interesting, so I am thinking of writing my own
solution. To solve the issue of being able to create an index page or filter
by categories, I'll use oq, a wrapper around jq that can parse YAML, to parse
the YAML metadata of each file and either generate the HTML directly or send
that as more YAML metadata to pandoc with a special template.
Using pandoc fixes the main problem that I currently have with my site: the
Markdown support is pretty poor and I find myself needing more and more out of
Markdown with each of my posts. I might end up relying on pandoc privately and
just sending its generated HTML to my blog's admin page until I do the rewrite,
so that I can avoid the poor PHP parser.
Comments
It took me a while to decide, but I chose to just not include comments at all in the next version of Brainshit. It will just be articles, and that's all. We got very few comments, half of those were spam, and replying was hard since there were no email notifications. If we want replies, we'll just publish our email addresses or other contact info. This makes a static site much easier to consider; the few solutions available for comments either require me to host another service that gets backed by a database somewhere, and the whole point was to remove the dependency on binary files, or require me and the readers to have GitHub accounts, which is not acceptable considering the fact that GitHub is an American company, owned by Microsoft, and makes questionable choices regarding its clients.
Article ideas
- Personal knowledge management
- A d7 throw costing 400 billion euros
- On imperfection: "Le talent d'Achille"
- How many points do we get by shooting every basketball in existence through a hoop at once?
- A conclusion on my use of the integrity report
- Projects that do not follow an integrity report or that do not seem to matter in some kind of productivity methodology aren't always negative distractions, they can also be just hobbies that we do for fun.