Email or username:

Password:

Forgot your password?
Top-level
Totally with Ю ⁂

@kepano @obsidian As I rely more and more on canvas, is there a plan to "open-source" canvas rendering? Markdown is there to last. But canvas is not a broad standard. While canvas files are text-based and therefore readable by any software able to, the correct rendering is an Obsidian specific functionality. What can we do to ensure future-proofness of canvas files beyond Obsidian?

EDIT: As it was pointed out, the canvas JSON format seems rather trivial to reimplement.

6 comments
Colman Reilly

@yuliyan @kepano @obsidian can't imagine it'd be more than a few days work to render them sensibly with something like GraphViz. Depends how much fidelity to the #obsidian display you needed, I suppose.

Colman Reilly

@yuliyan @kepano @obsidian I mean, this took me twenty minutes: a ruby script to read the JSON and render it into graphviz format. There are editors that can edit from that format.

A simple test obsidian canvas with four nodes and edges between them, showing the contents of the nodes, including one note node, rendered by Graphviz after reformatting into dot format.
Colman Reilly

@yuliyan @kepano @obsidian not really: the canvas format is pretty clear JSON, so translating it is easy. If/when #obsidian goes bad the install base is big enough that tools to translate to whatever replaces it will be available. Dataview is the biggest problem and I believe there’s a stand-alone version of that in the offing.

Totally with Ю ⁂

@Colman @kepano @obsidian I meant "amazing" as in realizing that the canvas JSON is already pretty resilient. Thank you for clarifying.

Regarding Dataview, I am not in the school of thought, that wants to see Obsidian have more database functionality that it does right now. The referencing between notes and folder structure enforce simplicity and resilience. Combined with tags and naming conventions it is already pretty powerful without dependency cost.

Totally with Ю ⁂

@Colman @kepano @obsidian I believe many things that Dataview does can be done by improving the core search functionality. When it comes to search, I. believe that computers in 50 years should easily be able to display search results based on human language queries. Data should be observable but does it need a technical database implementation other than search indexing?

Go Up