I kind of wish they would plant a stake in the ground and declare it feature complete and then go into maintenance mode. It's almost inevitable that they are going to keep adding features, give up control to VCs, and then fade away like Evernote.
This feature is neat, but it feels like a turning point. I believe it's the first betrayal of the it's-just-a-folder-of-markdown-files principle.
I agree that VC basically killed Evernote. Obsidian is completely user-supported, no investors, we're explicitly avoiding the VC route.
As far as features go, we're continuing down the path of a modular architecture. The core will continue to be as streamlined as possible. Canvas is like any other plugin, you can disable it. We think that flexibility is important, because not everyone thinks the same way. You should be able to create an environment that fits your way of thinking. However not everyone needs every feature, thus the modularity/extensibility approach.
Did the core of Obsidian have to be changed at all to support Canvas? If so, then I don't think it is like any other plugin.
If this were just another pluigin that you could download and use if you want, I really wouldn't care. It's the fact that it's considered part of the core product that gives me bad vibes. It feels like the project, even without VC money, is doing what almost every other successful project does - expand.
Do you think Obsidian will ever be considered "done"?
We did make some API changes that would make it easier for a third party to create a plugin like Canvas. But there was nothing changed in the core solely to support Canvas.
There were two under-the-hood changes required to properly facilitate Canvas:
1. Redoing "embeds." We decided it was important for cards to work the same way our inline embeds work. So we rewrote the embed system to be properly extensible and useable by plugins. Canvas is leveraging the same system that powers ![[embeds]]
2. Untangling our editor. Previously, editors were a construct constrained to a markdown view in Obsidian. We've also done some refactoring so that a "editor" can be used anywhere, and doesn't need to be backed by a file. We see this a another big win for plugins that want to build their own editing experience
We want to keep pushing what third party plugins can do on top of Obsidian, so implementing a feature like Canvas forces us to find the limitations that exist in the API.
The question of whether Obsidian will ever be "done" is a tough one because operating systems and user expectations are a shifting landscape. Our intention is that the writing and thinking you do inside of Obsidian can be future-proof for decades, even if Obsidian itself is no longer relevant. That's why we're focused on portable formats like Markdown. However we cannot know how operating systems will change, and what will be required for Obsidian to continue working well on macOS 30 or Windows 20. Similarly, we can't close ourself off to new UI paradigms like Canvas that open up new thinking modalities our users are asking for. We hope that the modular and flexible architecture of the app allows it to remain very performant regardless of what plugins a user has turned on.
You can represent a collection of nodes like that, but human readable/machine readable flow configuration, highlighting, plus composition seems like a tall ask, and JSON is a small extension that's mostly human-readable as it is.
This is adhering to an obvious crack showing in Obsidian as it is: how do you store a graph view configuration? Right now, per vault, you get one slot unless you draw in something like Juggl, which is... well it raises serious usability concerns.
It's just a plugin i believe, you can choose to not use it. There are _many_ community plugins that go well beyond just-markdown, i don't feel this is any different.
Remember, most of Obsidian is just plugins. Even core functionality. Which is a big reason i use Obsidian.
This feature is neat, but it feels like a turning point. I believe it's the first betrayal of the it's-just-a-folder-of-markdown-files principle.