khinsen: Just reading up on The Open Science Network and I saw that you are involved.
https://openscience.network/

h/t https://wedistribute.org/2024/03/bonfire-open-science/

Looks like a great initiative!

In reply to @indieterminacy:matrix.org

khinsen: Just reading up on The Open Science Network and I saw that you are involved.
https://openscience.network/

h/t https://wedistribute.org/2024/03/bonfire-open-science/

Looks like a great initiative!

Spent last night rebuilding my website (orionreed.com) and ended up, mostly successfully, building a PoC DOM -> Canvas translation. If you hit the button(s) at the top right you should see what I mean. I also wrote a little tweet thread that breaks it down a little more (https://twitter.com/OrionReedOne/status/1772271156373905465)

I think there's a lot of "low-hanging fruit" in UX-first interoperability. Where you get the pieces and the materials able to coexist and talk to each other, then you enrich them with more computation. I think canvases as an Information Substrate (in the sense coined by Michel Beaudouin-Lafon) have a pretty unique potential in moving towards a more integrateable HCI.

I don't mean that everything is spatial, or everything is whiteboard-esque, but that the properties they have as a medium for information are actually very well suited for integration of UX in a way that the DOM is not:

• DOM is problematic to produce concurrently without coordination, leading to almost all production being at the scale of a page/document/website
• DOM Layout is heavily restricted and not easily composed with independently-authored work
• DOM has a tiny vocabulary for topological relationships (I mean in the representation, not in JS) such as hierarchy and sibling, where canvases can be much more expressive.

These are half-thoughts, I think there's some cool potential here, just wanted to poke this group and see if any thoughts emerge.

In reply to @bkil:matrix.org
Up to now, my preconception about canvas was the opposite of accessibility. Is there prior art how to build a whole SPA solely on canvas but still make it as accessible as if you were using DOM?

One near-term approach to this would be to try and keep all key information in a DOM which can be readily accessed by existing accessibility tech. Longer term, the challenges are shared with many other domains: How do we do accessibility for graph-shaped information? Or for interactions that are not transactional or command-based? Or where you need to interact with multiple overlapping hierarchies of information (and so on).

Lot's of open questions and it will always be harder to work against ubiquitous web standards/norms/tech because it's had decades to develop things like accessibility.

In reply to @orion_reed:matrix.org
The challenge with SVG is that it's a standard for drawing, not holding semantic or topological information, so for example a graph is made of actual geometry, not relationships. I did actually reach out to Chris Lilley recently, the inventor of SVG, as I recently roped myself into organising an interoperability working group for canvas systems (canvasprotocol.org)

imho the best descriptor of canvases is as information substrates (https://hal.science/hal-01614273/document). They define constraints on the data, context for its interpretation, and a means of presentation and interaction.

Certainly not a standard, as there are many for semantic and topological information. But canvases are in a slightly richer category as a medium which is intractable and visualisable from the outset. The working group I mentioned (with folks like tldraw, stately.ai, sage-3 and others) is hoping to make some headway on open standards/interop for not just presentational details but relational and semantic information too.