"Go ... ergomonics"
"G# compiles straight to managed assemblies and runs on the modern .NET runtime"
These two are nothing close because it completely misses Go's ergonomics of compilation to portable static-linked binaries. For language constructs, maybe they have similar ergonomics, but the language is not only constructs. You have to ship your programs somehow, and this is where requirement to ship both binaries and the interpreter for them, sucks.
One of the things go makes it simple to do is help a reader understand which constructs belong to which package, since any use of external imports is prefixed by the package name.
In go, if I don’t know a constructs definition, i know exactly where to look at and find it.
When exploring a new language I won’t always setup an ide first. I just want to look at the documentation on my own. Heck, any language which requires ide or any mandatory tooling to work with it, is already handicapping a developer.
G# seems to copy go syntax but didn’t copy the things that go makes it easier to understand any go code.
Agree, and this is a huge bungle you can make in language design, that will affect compiler performance down the line significantly.
If you know which package each type comes from, you only need to check the exports of that package, otherwise its a local.
Otherwise you have to start compilation by building up a list of every package and its exports. This is particularly bad for incremental compilation, and starts to drag on large projects.
The saving grace of .NET in this case is that most .NET programs traditionally didn't have that many imports. You have mscorlib for the framework itself, which is almost all the basics, and maybe ASP.NET and a dozen other small packages.
But this doesn't have to be the case, and if you have thousands of small packages (because your app is huge), then your performance tanks.
While the language itself doesn't seem very interesting (which is perhaps the point). The idea of building a new language on top of the CLR runtime feels very wise. Instead of a new language suffering from a lack of ecosystem, you get everything else that's already built, even in other (CLR-based) languages. This does leave me wondering what the trade off is. Do you sacrifice any potential language features for CLR compatibility? Or provided you can get it compile, it will run?
Also, are there other language ecosystems with a similar capability?
The best AI models at this point are perfectly fine handling new languages and frameworks and stuff - as long as you can point the llm to some docs and some example code it's not gonna do noticeably worse than it would at another language.
Sure, but you expect of AI the ability to use the vast knowledge on languages and platforms that already exist out there. Just because they can be toughy to understand a new language won't make them proficient in it, able to save problems, etc.
These two are nothing close because it completely misses Go's ergonomics of compilation to portable static-linked binaries. For language constructs, maybe they have similar ergonomics, but the language is not only constructs. You have to ship your programs somehow, and this is where requirement to ship both binaries and the interpreter for them, sucks.
reply