It's a bit sad that the focus is only on Compose for Web (which is quite heavy and renders to a canvas), which isn't really ideal for many use cases, instead of Compose for HTML, which could've provided an alternative to real web frameworks, like Angular... Luckily there is some promising stuff from the community to hopefully bring something similar to WASM and maybe even get SSR working on the JVM as well, so you don't even need a nodejs server for the SSR.
Compose for html is limited. I did have fun using it, but you have the same limitations as with other frameworks: web is 3 different components, all of which have their own states, namely css, html and JS. You can't fully control everything, you are encapsulating and fighting the frameworks. For that you can already use react in kotlin. Compose for wasm would be awesome, as then you can actually reuse a lot of code and concepts from other platforms. This would be very difficult with html.
Can you use a HTML-backed multiplatform compose stack? As in, write once, run everywhere like with wasmJs + canvas, but instead of the canvas, it uses regular DOM elements?
@@SIMULATAN If you have the time to do a proper UI for all platforms or at least Web and non-Web, you can build a real web application with that and still don't need a second pattern and rely on a JavaScript library like react, but you can do it faster and more efficient in WASM for example (not with compose for HTML directly though as it has never been updated with support for WASM since that came out, only JS). And I count stuff like browser translation, being able to actually select text, better accessibility, power savings though browser native rendering and more as a win for something like compose for HTML compared to basically what is basically a remote desktop session into web assembly when using compose for web. Technically it has more in common with an interactive video that's live generated than an actual normal web application.
Rust is a great language, but it's hard to learn and understand and it takes much longer to write code. But if performance and reliability is critical it's awesome.
Targeting WASM opens the door for so many more opportunities to use Kotlin in novel ways. Keep it up!
It's a bit sad that the focus is only on Compose for Web (which is quite heavy and renders to a canvas), which isn't really ideal for many use cases, instead of Compose for HTML, which could've provided an alternative to real web frameworks, like Angular... Luckily there is some promising stuff from the community to hopefully bring something similar to WASM and maybe even get SSR working on the JVM as well, so you don't even need a nodejs server for the SSR.
Compose for html is limited. I did have fun using it, but you have the same limitations as with other frameworks: web is 3 different components, all of which have their own states, namely css, html and JS. You can't fully control everything, you are encapsulating and fighting the frameworks. For that you can already use react in kotlin. Compose for wasm would be awesome, as then you can actually reuse a lot of code and concepts from other platforms. This would be very difficult with html.
Can you use a HTML-backed multiplatform compose stack? As in, write once, run everywhere like with wasmJs + canvas, but instead of the canvas, it uses regular DOM elements?
@@SIMULATAN not multi platform, but there are libraries for that across js and wasm
@@cromefire_i see, thanks!
@@SIMULATAN If you have the time to do a proper UI for all platforms or at least Web and non-Web, you can build a real web application with that and still don't need a second pattern and rely on a JavaScript library like react, but you can do it faster and more efficient in WASM for example (not with compose for HTML directly though as it has never been updated with support for WASM since that came out, only JS).
And I count stuff like browser translation, being able to actually select text, better accessibility, power savings though browser native rendering and more as a win for something like compose for HTML compared to basically what is basically a remote desktop session into web assembly when using compose for web. Technically it has more in common with an interactive video that's live generated than an actual normal web application.
What are the Kotlin team's future plans for supporting other WASM runtimes (incl embedded runtimes)?
I really miss Andrey Breslav, he was such an elegant speaker, I find it difficult to follow Zalim
In the meantime I write wasm production ready code in Rust where I don't need garbage collection 😁
Unfortunately I do not have the energy or time to learn Rust
Rust is a great language, but it's hard to learn and understand and it takes much longer to write code. But if performance and reliability is critical it's awesome.