Good questions. WASM at this point seems more a locus of JS hate than anything practical. To create something cool, novel, and useful should be the goal, not avoid or use a certain language.
There's a pretty stark difference between those working in and with WASM, and people commenting on the sidelines. Almost everyone involved with WASM itself sees it as complimentary, and explicitly is not trying to "kill javascript." Random commenters on the internet do say this a lot though.
WASM isn’t very practical yet because it isn’t fully formed. There isn’t support for DOM access or parallelism, never mind the ecosystem. It’s a really ambitious project, and writing it off as “locus of JS hate” because it hasn’t completely changed the game in such a short time frame is really short sighted. It would be a fairer test to give it 5 years.
I'm not writing it off. I think it's a good idea to bring more options to the web. I tried it last year, but as you state and I agree, it's not ready yet.
In my experience in SF I encounter many never-JSers. Being a JS fan, for certain things, I don't get the hate. So maybe I feel threatened in a way, but I'm not going around wishing for the death of python or Haskell. That's all. I am reading random internet commenters and getting slightly upset. That's on me.
I wouldn't say it's hate towards JS but I think the evolution of JS vs the evolution of web and its use cases have not increased at the same velocity, latter being faster. JS has had to have a lot of scaffolding added to increase its use but WASM now comes in with the understanding of the modern requirements. However, I don't think it's either one or the other, it'll depend on the use-case.
I can't imagine JS will go anywhere anytime soon, especially since it still needs to be called from DOM manipulation,but the ability to code a full stack in a single non-JS language and run it at near-native performance is going to be very useful as more apps become web based and complex. Even from a skills point of view, you can now have backend and front-end developers skillsets are slightly more merged so easier to swap around resources.
You’ll be happy next time you npm import something time sensitive that requires computation, and that you can just use from js for free.
Think about how you would do ML with just python. That’d be insane. And thankfully the packages you need are implemented in c or cuda with python bindings. 99% of ml people never see the insides of np, sklearn or tf. But you’re glad these aren’t implemented in python.
Similarly, you’re glad for webgl if you need to do lots of parallelizable computations (cfd, ml, 3d). But glsl was never bashing at js. Just that using js for these task isnt feasible.
Similarly, now you can use fast code from js. Essentially, you have a bunch of matrices use webgl, otherwise wasm. And orchestrate everything from js. Like ml peeps do with python, c and cuda.
steveklabnik|6 years ago
pteredactyl|6 years ago
weberc2|6 years ago
pteredactyl|6 years ago
In my experience in SF I encounter many never-JSers. Being a JS fan, for certain things, I don't get the hate. So maybe I feel threatened in a way, but I'm not going around wishing for the death of python or Haskell. That's all. I am reading random internet commenters and getting slightly upset. That's on me.
keyP|6 years ago
I can't imagine JS will go anywhere anytime soon, especially since it still needs to be called from DOM manipulation,but the ability to code a full stack in a single non-JS language and run it at near-native performance is going to be very useful as more apps become web based and complex. Even from a skills point of view, you can now have backend and front-end developers skillsets are slightly more merged so easier to swap around resources.
globuous|6 years ago
Think about how you would do ML with just python. That’d be insane. And thankfully the packages you need are implemented in c or cuda with python bindings. 99% of ml people never see the insides of np, sklearn or tf. But you’re glad these aren’t implemented in python.
Similarly, you’re glad for webgl if you need to do lots of parallelizable computations (cfd, ml, 3d). But glsl was never bashing at js. Just that using js for these task isnt feasible.
Similarly, now you can use fast code from js. Essentially, you have a bunch of matrices use webgl, otherwise wasm. And orchestrate everything from js. Like ml peeps do with python, c and cuda.
For the better or the worse...
galaxyLogic|6 years ago
thefounder|6 years ago