Telegram Messenger has frameworks for most known languages, which makes easier for anyone to integrate with them. I started with Golang and soon found that those frameworks are not up to date, not to mention my experience testing on Golang is also mixed due to how their testing tool works. The natural runner-up was JS, which I'm ditching in favor of TS to make a strongly typed code, proper tests and documentation for broader usage. TypeScript allows fast prototyping and can prevent problems during code phase, given that your IDE of choice has support for a language server, and build phase. Pairing it with lint tools also allows honing code before it even hits the repositories.

1. Type safety and inferred types

Go is type safe by default, which allows you to right more reliable code and have better developer tooling, plus with the := operator, you can initialize a variable without having to define its type because it automatically gets its type from the initial value.

2. Performance

There isn't much to be said here, but on most counts go beats both Python and Node.js on performance.

3. Documentation

I'm not talking about the Go language itself, although it does have good docs. I'm talking about Go's auto generated documentation tool, which allows people to document their packages easily and works amazingly with Go's type system.

4. Compiles to binary

If you are making a local program for somebody and they don't want to download the Go compiler, you can make Go into a native binary.

5. Built for the web

Go has built in Http libraries to rival Express.js and has a HTML/Text templating system.

6. Great Concurrency

Go utilizes Goroutines to help developers utilize multiple threads easily.

Conclusion

Go is an excellent choice for any system code, especially http networking and web backends.

C# and .Net were obvious choices for us at LiveTiles given our investment in the Microsoft ecosystem. It enabled us to harness of the .Net framework to build ASP.Net MVC, WebAPI, and Serverless applications very easily. Coupled with the high productivity of Visual Studio, it's the native tongue of Microsoft technology.

Node.js has been growing in popularity, and the ability to access the global pool of Javascript developers is great. There is a decreased amount of effort for people to work across the frontend and backend, and the language itself is easy and works well for many common use cases.

Go was the other serious candidate, but it just hasn't been implemented in as many Production systems yet, and the best Go engineers I've known have been hackers, whereas we're building a robust analytics platform that requires more caution. Type safety is easily added with TypeScript, and NPM is awesomely handy.

When developing a new blockchain, we as a team chose Go lang over Java and other candidates, due to Go being (a) natively suited to concurrency - there are primitives in the language itself (goroutines, channels) that really help with reasoning about concurrency (b) super fast - build time, running, testing are all much faster that Java, this gives a far superior developer experience (c) shorter and stricter than Java - code is much shorter (less verbose), and there is usually one good way to do things, and even the code formatter that is bundled with Go is very opinionated - over a short time this makes reading other people's code far smoother than having to deal with different styles.

You should be aware that Go presently (v1.13) lacks Generics.

I chose Golang as a language to write Tango because it's super easy to get started with. I also considered Rust, but learning curve of it is much higher than in Golang. I felt like I would need to spend an endless amount of time to even get the hello world app working in Rust. While easy to learn, Golang still shows good performance, multithreading out of the box and fun to implement.

I also could choose PHP and create a phar-based tool, but I was not sure that it would be a good choice as I want to scale to be able to process Gbs of access log data

We decided to use python to write our ETLs and import them into metabase via a lambda. Before python we tried using Go, but overall go was way more verbose than Python when writing the ETLs. Go also had some issues managing memory when using the S3 upload manager library. This was a deal breaker for us that made us switch to Python.

In the end the solution was much cleaner and maintainable.

This language, even in early dev stages is to put it simply, fantastic! It is small, fast, and types a lot like go. It feels complete even though coming out less than a year ago in first early stages. I love it, it works anywhere and everywhere plus making binaries and GUI applications is just super easy!

A number of years ago; I had done python for a long time prior to learning about Go. Most of what I wrote was system-like things and web-things in python, and I got tired of running into the lack-of-a-type-system problems that python gave me. I wanted to switch to a compiled, strongly-typed system that wasn't C/C++ (been there, done that, got the "shoot yourself in the foot" t-shirt). I looked into both Rust/Go, and for what I wanted to do (system/web) stuff ... at the time, Go was the strongest candidate, so I switched and never went back. Recently I started to re-look at Rust for system things, but for anything I do that I have to touch the web with, it will be Go from now on.

As a personal research project I wanted to add post-quantum crypto KEM (key encapsulation) algorithms and new symmetric crypto session algorithms to openssh. I found the openssh code and its channel/context management extremely complex.

Concurrently, I was learning Go. It occurred to me that Go's excellent standard library, including crypto libraries, plus its much safer memory model and string/buffer handling would be better suited to a secure remote shell solution. So I started from scratch, writing a clean-room Go-based solution, without regard for ssh compatibility. Interactive and token-based login, secure copy and tunnels.

Of course, it needs a proper security audit for side channel attacks, protocol vulnerabilities and so on -- but I was impressed by how much simpler a client-server application with crypto and complex terminal handling was in Go.

$ sloc openssh-portable 
  Languages  Files    Code  Comment  Blank   Total  CodeLns
      Total    502  112982    14327  15705  143014   100.0%
          C    389  105938    13349  14416  133703    93.5%
      Shell     92    6118      937   1129    8184     5.7%
       Make     16     468       37    131     636     0.4%
        AWK      1     363        0      7     370     0.3%
        C++      3      79        4     18     101     0.1%
       Conf      1      16        0      4      20     0.0%
$ sloc xs
  Languages  Files  Code  Comment  Blank  Total  CodeLns
      Total     34  3658     1231    655   5544   100.0%
         Go     19  3230     1199    507   4936    89.0%
   Markdown      2   181        0     76    257     4.6%
       Make      7   148        4     50    202     3.6%
       YAML      1    39        0      5     44     0.8%
       Text      1    30        0      7     37     0.7%
     Modula      1    16        0      2     18     0.3%
      Shell      3    14       28      8     50     0.9%

https://gogs.blitter.com/RLabs/xs