Also Zettlr
- self hosted
- works offline (mostly)
- "just md" BUT
- scriptable or extendable by lua, rendered in page, eg `${1 + 1}` outputs `2`, but you can do a lot more, such as query pages and tags with a LINQ type query interface.
People use chatbot on the mobiles - way more convenient.
You can both read/write notes through the chat.
So that your files and tools can grow together, fully under your ownership, through the ages.
The app can be easily tweaked for your own needs via an LLM - code is optimized for that.
P.S. And Golang seems to be great fit for this kind of software.
docker + php-fpm + php7 + larvel + nginx + redis + cron + worker + certbot
Server after the rewrite to Golang:
server, a 15MB no-dependencies binary that has everything.
Like, I should be able to open it even after a few years, and do some fixes or add some features.
Go's ecosystem seems to share this mindset.
I believe I put too much time into it during all those years, but I don't regret it. Because I use the project on daily basis.
It seems like software in AI-era should be distributed open source.
So that anyone could tweak it however he wants. Not though buggy plugins system.
For the first time, I put a sponsorship button. Will see if it works.
The subscription based platform with automatic cloud hosting and other quality of life features, whatever those are depending on the app.
Although there's a bunch of 100% open source projects and developers that get enough donations to make it their full time job just off of that. Not that it's the way to go if you want to get rich, but it's still very much a real thing.
In my experience, if the dev wishes to be compensated in dollars, they also sell a commercial license, cloud services, etc.
I have a problem, I spend a few days building a tool that solves the problem, it works pretty well for me, and I release it to let others get value from it. They make tweaks to it, perhaps improve it, and I get value from those enhancements and bugfixes.
That makes it easy for AI to be trained on it.
That's the point of open source, sharing the knowledge.
We'll all make the same shit over and over if noone shares.
But if we all share, then the only thing left to make is the unknown.
That was true before the "AI era" as well.
Just now, any regular user can clone the repo and ask an LLM to tune it to his needs.
That's why I will always hammer on open standards and federation.
When I read "an alternative", I assumed feature-parity and API compatibility. But what I found out was entirely different and much more interesting.
I'll give it a try, thanks for sharing your year-old work!
When I read “alternative” I immediately had a rant in my head about people calling things alternatives that are not.
I looked hoping feature-parity isn't a goal. Obsidian is there for the people who want it. Do something more interesting than that
Thanks for a good observation! Indeed, I don't position it as Obsidian alternative. I don't know a better pitch for it just yet.
For me that's something about: simplicity, lazy flow of adding things, readiness to use out of the box.
To focus on what works, and not what is fancy.
"TextBundle brings convenience back - by bundling the Markdown text and all referenced images into a single file."
We had those already more than a decade ago. Personally, I fondly remember Mou.
Obsidian has heavy Electron vibes, and Files.md is several steps more into the wrong direction.
The name is also bad. It feels like it was chosen because someone already had the domain.
There is no better interface for text than a terminal, and we are in the golden age. Despite being extremely powerful, this setup will run on resource constrained machines.
It's a personal choice that cannot be imposed on everyone. Not everyone is a developer.
krthr•1h ago
zakirullin•1h ago
dewey•1h ago
zakirullin•54m ago
A lot of us built knowledge bases, and we enjoyed it all quite a bit.
snjnlsn•25m ago