0:00 start 0:17 outline 1:40 goals and scope 3:07 DOM or not DOM 8:55 Layered UI 22:08 The four tribes of declarative UI 39:35 Xilem overview 44:47 xilem_core: native + Web (demo!) 1:08:37 Vello rendering 1:17:10 Hypotheses to test 1:21:05 Next steps
One thing I would have liked in the justification section is what "slow" means. I think it would be useful to have an example of an experience someone might like to build but is not possible or is very difficult because of performance limitations on the web
I work building software for trading stocks. We would like to update the market data more frequently, but we can't because the DOM and current web reactive layer is too slow.
One distinction I’ve heard of is performance: you don’t necessarily want to hammer your cpu/gpu in a UI setting, but might be more ok with it in a game. I think the distinction is a bit murky though, there is a lot of overlap between the two.
Great talk, Raph! It seems like Xilem is covering a lot of bases when people ask questions like "But what about XYZ?". I'm hoping this becomes the de facto way to write UIs in Rust as we sorely need a better UI story.