why we need dotnet.wasm

[sajjadarashhh]

why we need dotnet.wasm?
while browser that have owned runtime and we can execute wasm codes in that
if we compile c# modules directly to webassembly modules then why we need dotnet.wasm?

[hez2010]

Because currently we don't compile C# to webassembly modules directly. Instead we are using dotnet.wasm to interpret IL code.
You might find something interesting here: https://github.com/dotnet/runtimelab/tree/feature/NativeAOT-LLVM, which is the on-going work of compiling C# to webassembly directly with NativeAOT.

[Clockwork-Muse]

"But it's faster" doesn't actually mean "my code is currently too slow".

[sajjadarashhh]

@Clockwork-Muse in global benchmarking that says dotnet runtime is too slow in browser

[Clockwork-Muse]

@sajjadarashhh -
That's not usually what benchmarks actually measure. They normally give "this task takes this long/this many resources" answers, and then you have to decide if that makes sense for your usecase.

For instance, a month ago, I developed a small proof-of-concept that received events from an MQTT broker and displayed a bunch of stats on the screen. I had a poor implementation due to the rush, but it worked well enough for the PoC.

[sajjadarashhh]

Size is very big 😐
We dont have critical issue in runtime we have many problem with startup time and first time load time
For example if you see mudblazor site you dont wana to open a pages in new tab for .net runtime load time... 😑🤦

[hez2010]

Slow speed + large size = no one is willing to use it in production (especially those websites which will be accessed by public users).

• Slow speed: blazor costs a lot of time to initialize to load the runtime and render the first page, even if you have all those wasm files cached. This is very easy to observe on a low-performance endpoint such as smart phones.
• Large size: spend more money for delivering contents from a CDN because it requires more network traffic.

The WebSocket based interactive server-side rendering doesn't resolve anything as it only moves the workload from clients to servers, which will cost more compute resources, and it also makes it impossible to scale out/restart the server seamlessly because all states will be lost if the server down.

Without NativeAOT-LLVM, blazor is only suitable to develop projects for internal use (i.e. those projects which are not intended to be accessed by public users) for both performance and size reason.