So cool. Wanted to build something using WASM for like forever. Always hesitated because I felt the setup and build-processes aren't streamlined at all. Wasmer seems promising. Can't wait to try it. The WASM book and the MDN docs didn't give me the push to go on with WASM. This video did. Thanks a lot.
It's not enough that they are really talented developers, it's the fact that they work together as a team or just helpful contributors, most of the time! We often underestimate that kind of human relationship: working together towards a better future. We have a lot to learn from these kinds of interactions and their results in the real world, be it software related, NGO related, or just people in a local community building a better community for everyone.
I'm the contributor of cross-lang-and-cross-platform project and I'm excited to see more people tackling the problem of cross-language interoperability!
I'd be curious what the common known pitfalls of that are, and how to avoid them. Like the xkcd "one standard to rule them all" becoming "now we have yet another standard." As someone else said, we don't need yet another JVM or DCOM or CORBA. But if there's a way to make bi-directional FFI convenient and architecture independent, I'm all for it.
I really like the idea, and believe that this can be used in the future by smaller/newer languages to have strong ecosystems, given they implement wasm integration. The only problem that I see right now, is that you don't have LSP for the WAI bindings on Rust for e.g, so this makes developing anything more complex kinda annoying.
Are wai files standard format or is it something by wasmer? with how webassembly is, I'm scared it might turn into a mess of an ecosystem like javascript
I am quite surprised the .wai file is not automatically generated based on the Rust code. I realise this might be a problem for a library written in a dynamically typed language, but it would be much easier if the file were generated where possible.
The Rust code is generated from the wit file, which is better IMO, because you define the protocol first, and then the implementation. But it would be cool if it would be possible the other way too.
also the client code need to specify the funtion to call as a string, that seems weird, specially in strong typed languages like Java or C#. if they solve that, I'll take a look at this product, meanwhile i will evade it.
Great video! It seems to have plenty of use cases for this. Do you know if we could also use the webassembly sandbox mecanisim with this approach? I could imagine a plugin system with restricted rights.
I think the only drawback would be that WASM may be slightly less performant than the equivalent logic running on bare metal (machine code). This will be a non-issue for some use cases, and a complete deal breaker for others.
This could be amazing. I love using Rust built items, but I'm not a big fan of coding in it, yet. Right now I'm still heavily coding in Go and their dataframe handling package is unmaintained. I've been eyeing polars (and a lot of other crates) and have been wishing to use those in Go. I know the real solution is to just use Rust, but that's not always an option to rewrite everything. Maybe this will be the start for utilizing more Rust under the hood of other languages?
You don't necessarily have to write everything if you use the web services architecture; I know it's kind of a buzzword, but part of why it's great is that you can implement your backend in multiple languages. You can keep all of your non-performance-critical code in Go and fallback to Rust when you really need it.
Your entire app does NOT need to use one language of the other. "Interprocess Comunicaition" and "Foreign Function Interface" might be good search terms as general ideas to interoperate Rust and Go (Other than just using wasm, of course)
Random off topic question. How does he have that shell command input on the bottom in emacs? I want to get the same in Neovim. I don't want to type :! every time.
you get that "out of the box" with DOOM - I press M-! to get it. One fewer keystroke than nvim I guess. I'm also curious about a quicker alternative in neovim.
Oh man! I undestimated WASM, now i realized that you can send "binary" module but not have to compile it for my arch.... devastating. Is any standalone executor/sandbox for wasm exist excluding browsers and node/deno etc. ?
is this an implementation of the new "Component Model" in WASI? Great video, btw - thanks for sharing - packed with good info. When a language with a runtime publishes to WA, will that include a WA coded runtime for that language (like C#)? thx IE, will it perform like a WA module coded in Rust, or more like a garbage collected language.
I have a question. I’m currently working on a personal project and I’m wondering if there’s a way to define 2 functions with same name but different argument types and return types in the wai file? Similar to what we can do with traits (think function overloading)
For anyone trying to recreate this on their own: *WAPM is deprecated*, and so `cargo wapm` has been replaced by `cargo wasmer`. This is how the metadata configuration looks like in `Cargo.toml`: [package.metadata.wasmer] namespace = "" # same as your wasmerio account handle abi = "wasi" # choose "wasi" for both javascript and python bindings bindings = { wai-version = "0.1.0", exports = "" } # should not contain `_` underscores
nope you can have what are essentially "classes" as parameters as well - or "records" in WAI terms. I go a bit more in depth into WAI and its capabilities toward the end of the video.
I've searched their main site, their docs site, and glanced at their github, and I still can't figure out what languages they support, which is MINDBOGGLING. D:
There used to be such a list, I think they recently did a major revamp of the website (and merged wapm.io into wasmer.io). It seems like there is a list of the major languages here wasmer.io/products/runtime but I'm also not sure if the comprehensive list exists somewhere. Is there a particular language you're curious about?
What would be cool if you could use the generated WAPM package without publishing it to the registry. That's going to be particularly useful for Monorepos and private code base for people who do not want to publish everything to public WAPM registry.
Ok, I like the idea of webassemblies but there is some problem because I can't write libraries specifically for webassemblies, this problem is a problem with rust collections. If anyone knows how to solve this problem for example HashMap answer me
Someone should re-write all of the Python libraries into Rust. That way, a collection of universal libraries, written in Rust, could be called from any other language that a programmer may prefer to use. Is there any software that will assist in converting C/C++ source code into Rust source code?
I don't like that the wai file has lowercase everywhere, but the code generated has camel case. Like, why is the "crab" record generated as "Crab" in rust?
Actually, the WAI interface is not supported by the Bytecode Alliance, which is defining the officiel WASI protocols. WAI is a fork of wit-bindgen, because wasmer tried to integrate bindings to wasmer into wit, and got rejected, because those 3rd party bindings do not belong in the source project. But the Wasmer CEO got butthurt, and make his own fork from WIT, which he named WAI. You all should rely on the Compoment Model and WIT, worked on by the original WebAssembly designers : The Bytecode Alliance.
I'm aware of the WAI origin story, though I'm not sure anybody got "butthurt". I don't think it makes sense to stick with something solely because it's maintained by the original creators, if that was the case we'd all be using vi or ed instead of Neovim. If you can enumerate the advantages one would get by sticking with WIT, that might be more constructive - for example do they have an equivalent of WAPM?
getting this error on cargo wapm: Error: 'AnonymousUser' object has no attribute 'is_email_validated' Error: Unable to publish "universal_crypto" my email in wasmer is verified...
I had no idea wasmer could do that! That's awesome
it's a fairly recently introduced capability, so it's understandable. I'm pretty excited about it too
So cool. Wanted to build something using WASM for like forever. Always hesitated because I felt the setup and build-processes aren't streamlined at all. Wasmer seems promising. Can't wait to try it. The WASM book and the MDN docs didn't give me the push to go on with WASM. This video did. Thanks a lot.
nice, really glad you got something out of the video! if you're willing, definitely let us know what you wind up building with it
Sure.
Great video! Wonder how it treats the types for TypeScript, does it emit `.d.ts` files as well?
Wow this is really cool. Also I need to learn more about web assembly, definitely checking out that video!
I agree it's pretty exciting!
It's not enough that they are really talented developers, it's the fact that they work together as a team or just helpful contributors, most of the time! We often underestimate that kind of human relationship: working together towards a better future. We have a lot to learn from these kinds of interactions and their results in the real world, be it software related, NGO related, or just people in a local community building a better community for everyone.
Most NGOs are sponsored by some government with a clear but hidden agenda, so that should probably be excluded.
That's exactly what I needed for the library project I'm currently building
I'm the contributor of cross-lang-and-cross-platform project and I'm excited to see more people tackling the problem of cross-language interoperability!
got your mail, Thanks Dimitrii If you would like to connect dm me in community slack
Great! Rust + WASM = super powers
agree 💯
i always dreamt of such things, using webassembly as common between language given every language is trying to integrate with webassembly in some way.
yeah it seems perfectly positioned to fill that need!
I'd be curious what the common known pitfalls of that are, and how to avoid them. Like the xkcd "one standard to rule them all" becoming "now we have yet another standard." As someone else said, we don't need yet another JVM or DCOM or CORBA. But if there's a way to make bi-directional FFI convenient and architecture independent, I'm all for it.
I really like the idea, and believe that this can be used in the future by smaller/newer languages to have strong ecosystems, given they implement wasm integration. The only problem that I see right now, is that you don't have LSP for the WAI bindings on Rust for e.g, so this makes developing anything more complex kinda annoying.
moving your face to show the code which otherwise woulda blocked the code surprised me and is much appreciated!
You get a reference to self when using a resource automatically? That is *so good*! Can't wait to try it out.
😎
Are wai files standard format or is it something by wasmer? with how webassembly is, I'm scared it might turn into a mess of an ecosystem like javascript
Very good product. Congratulations guys. Kisses from Romania looking forward to implement your awesome product.
I agree it shows great promise!
which font are you using ? looks good to be honest
If you look at the lowercase 'i', they are different in the terminal and in the editor.
In emacs it is Monaco, in VSCode i *think* it is Menlo
Damn, the "WASM Interface Types" dream is true. I will have to see if I can use this instead of fp-bindgen.
its worth a try!
I am quite surprised the .wai file is not automatically generated based on the Rust code.
I realise this might be a problem for a library written in a dynamically typed language, but it would be much easier if the file were generated where possible.
The Rust code is generated from the wit file, which is better IMO, because you define the protocol first, and then the implementation.
But it would be cool if it would be possible the other way too.
also the client code need to specify the funtion to call as a string, that seems weird, specially in strong typed languages like Java or C#. if they solve that, I'll take a look at this product, meanwhile i will evade it.
Great video! It seems to have plenty of use cases for this. Do you know if we could also use the webassembly sandbox mecanisim with this approach? I could imagine a plugin system with restricted rights.
this is awesome!! I wonder if it's possible to generate a Node package locally/without publishing it.
I found out it is, by using the wasmer-pack crate
okay, i completed watching the video. now i want know about the drawbacks or limitations of this.
I think the only drawback would be that WASM may be slightly less performant than the equivalent logic running on bare metal (machine code). This will be a non-issue for some use cases, and a complete deal breaker for others.
Does autocompletion work for these libraries? Does the lsp know that the uc object has the method hash with it?
This could be amazing. I love using Rust built items, but I'm not a big fan of coding in it, yet. Right now I'm still heavily coding in Go and their dataframe handling package is unmaintained. I've been eyeing polars (and a lot of other crates) and have been wishing to use those in Go. I know the real solution is to just use Rust, but that's not always an option to rewrite everything. Maybe this will be the start for utilizing more Rust under the hood of other languages?
You don't necessarily have to write everything if you use the web services architecture; I know it's kind of a buzzword, but part of why it's great is that you can implement your backend in multiple languages. You can keep all of your non-performance-critical code in Go and fallback to Rust when you really need it.
Your entire app does NOT need to use one language of the other. "Interprocess Comunicaition" and "Foreign Function Interface" might be good search terms as general ideas to interoperate Rust and Go (Other than just using wasm, of course)
Random off topic question. How does he have that shell command input on the bottom in emacs? I want to get the same in Neovim. I don't want to type :! every time.
you get that "out of the box" with DOOM - I press M-! to get it. One fewer keystroke than nvim I guess. I'm also curious about a quicker alternative in neovim.
rust compile is so slow even for this kind of project. anyway to speed up, caching or just direct linking binaries?
Is it possible to develop universal client and does it work with reqwest?
Oh man! I undestimated WASM, now i realized that you can send "binary" module but not have to compile it for my arch.... devastating.
Is any standalone executor/sandbox for wasm exist excluding browsers and node/deno etc. ?
so is it running rust code or is it generating code in the language you use it?
Informative 👏🏻 btw what font are you using it looks pretty cool 🙂
Monaco!
@@codetothemoon thank you!!
6:09 scared the heck out of me lol
How do i use wasmer as a dependency and include a wasm file to execute a function off that?
Ive tried several ways and just get a wasi error .. :|
This is so cool, I need to try it
it's worth giving a try!
is this an implementation of the new "Component Model" in WASI?
Great video, btw - thanks for sharing - packed with good info.
When a language with a runtime publishes to WA, will that include a WA coded runtime for that language (like C#)? thx
IE, will it perform like a WA module coded in Rust, or more like a garbage collected language.
I have a question.
I’m currently working on a personal project and I’m wondering if there’s a way to define 2 functions with same name but different argument types and return types in the wai file?
Similar to what we can do with traits (think function overloading)
For anyone trying to recreate this on their own:
*WAPM is deprecated*, and so `cargo wapm` has been replaced by `cargo wasmer`.
This is how the metadata configuration looks like in `Cargo.toml`:
[package.metadata.wasmer]
namespace = "" # same as your wasmerio account handle
abi = "wasi" # choose "wasi" for both javascript and python bindings
bindings = { wai-version = "0.1.0", exports = "" } # should not contain `_` underscores
When using records, how do you instantiate an instance of the record type in the language using the library?
Great video! Do the wasmer functions only allow primitive types as parameters?
nope you can have what are essentially "classes" as parameters as well - or "records" in WAI terms. I go a bit more in depth into WAI and its capabilities toward the end of the video.
I've searched their main site, their docs site, and glanced at their github, and I still can't figure out what languages they support, which is MINDBOGGLING. D:
There used to be such a list, I think they recently did a major revamp of the website (and merged wapm.io into wasmer.io). It seems like there is a list of the major languages here wasmer.io/products/runtime but I'm also not sure if the comprehensive list exists somewhere. Is there a particular language you're curious about?
Great content as always.
The font he is using is 'Monaco'
yep!
"enough brainpower to center a or quit vim" - good one 😂
thanks 😎
What would be cool if you could use the generated WAPM package without publishing it to the registry. That's going to be particularly useful for Monorepos and private code base for people who do not want to publish everything to public WAPM registry.
This is amazing, thanks for sharing
thanks for watching!
I wonder how the performance stacks up.
as compared to native machine code? I'd be interested to see that as well, I'd be surprised if somebody hasn't done it yet
Ok, I like the idea of webassemblies but there is some problem because I can't write libraries specifically for webassemblies, this problem is a problem with rust collections. If anyone knows how to solve this problem for example HashMap answer me
This will be the future
I'm still waiting for WASM's obfuscation to be used maliciously. It feels like WASM has trouble ahead.
Very cool!
thanks!
In the interest of assuaging my anxiety, you cycled all those credentials and tokens before publishing this video, right?
I did. Thanks for checking haha 😎
@@codetothemoon Whew. Thank you.
this means we can stable host COBOL applications remotely now
We can replace old iron now.
Zig vs rust Can you please make a video ?
In my experience wasm is considerably slower than native rust
interesting - how much slower?
👍Really nice👍!
Someone should re-write all of the Python libraries into Rust. That way, a collection of universal libraries, written in Rust, could be called from any other language that a programmer may prefer to use.
Is there any software that will assist in converting C/C++ source code into Rust source code?
wow
I was pretty impressed when I first learned about this!
I don't like that the wai file has lowercase everywhere, but the code generated has camel case.
Like, why is the "crab" record generated as "Crab" in rust?
Awesome, more malware libraries))
why limit yourself and make your malware consumable from only one language? 😉
@@codetothemoon you dirty boy!!!!!)) Love your humour and channel, subscribed already!)))
This is legit
I agree!
As soon as you opened vscode I died inside
I'll never understand the hate for vscode, no other general purpose editor even comes close to the features and extensibility of vscode.
This, is Elon Zuck.
This, is Mark Musk
This is basically CORBA 2.0. It failed once, I wonder if it will fail again.
Actually, the WAI interface is not supported by the Bytecode Alliance, which is defining the officiel WASI protocols. WAI is a fork of wit-bindgen, because wasmer tried to integrate bindings to wasmer into wit, and got rejected, because those 3rd party bindings do not belong in the source project. But the Wasmer CEO got butthurt, and make his own fork from WIT, which he named WAI.
You all should rely on the Compoment Model and WIT, worked on by the original WebAssembly designers : The Bytecode Alliance.
I'm aware of the WAI origin story, though I'm not sure anybody got "butthurt". I don't think it makes sense to stick with something solely because it's maintained by the original creators, if that was the case we'd all be using vi or ed instead of Neovim. If you can enumerate the advantages one would get by sticking with WIT, that might be more constructive - for example do they have an equivalent of WAPM?
It's pretty much expected that any library these days is in Typescript.
I think there are many use cases for which TS is not a good fit 😎
You are mistaken.
If only life also so simple 😂
sure, let's reinvent x86 platform, as a universal platform so we can run it on x86 platform. Now that's out of the box thinking
"Enoom"
FIRST
🥇
@@codetothemoon mom i finally did it
i'm up now.
@@skyline00069 🌄🥇
Postgres on the thumbnail tho
No bro, we don't want to use rust, C++ is fine.
getting this error on cargo wapm:
Error: 'AnonymousUser' object has no attribute 'is_email_validated'
Error: Unable to publish "universal_crypto"
my email in wasmer is verified...
Did you run the wasmer login command?