Do a benchmark on a production workload on a real application and we'll talk. Pretending like this project with 360 github stars is somehow authoritative for every language is silly.

Doesn't address the issue of how the concurrency was set up for these synthetic benchmarks, which again bear very little resemblance to a real web application task. Javascript has a different concurrency model and again, unrealistic compared to how a real web browser would run this.

The other thing about this particular set of benchmarks is that it's actually selecting for languages with good parallel threading support and concurrency in general, which has been a weak point of Ruby until recently. I don't know what threading approach they used but there are newer approaches such as Fibers instead of Threads, or the Ractor pattern. They also left elixir and erlang off the list. Elixir is a language with Ruby-like semantics but the concurrency of erlang. I'd be interested to see how it compares to JS.

These synthetic benchmarks that timed out probably aren't configured properly. For the ones that did run, ruby got decent performance considering they are using truffleruby which requires extensive tuning to take advantage of the JIT, and it likely isn't set up to do that, tweaking gc, and showing high memory usage. On a cursory look at their implementation https://github.com/hanabi1224/... it could certainly be improved to use hashes instead of recursively creating objects. The reality is that you aren't going to be computationally bound for common web application tasks. Synthetic benchmarks really just show that ruby is perfectly fine. I'm sure a knowledgeable ruby core contributor could show some significant performance gains in the ruby results here and probably the same could be said for python and other similar languages.

Ruby is a great replacement for Perl and PHP and adds many modern features for scalability. If you are productive in Perl and PHP and want to upgrade to a more modern stack, Ruby is a good choice alongside Python/Django, or Elixir/Phoenix

Nowadays I use asdf but you can solve this easily with rvm or rbenv. Which should probably be part of the standard ruby package management distribution, just as nvm is needed to use nodejs properly. If you try to run with a system library things are hard. Same thing with using the system python without virtualenv or anaconda or whatever.

Tell that to multiple large companies such as Shopify.

Concerns about rails performance are vastly overstated and mostly very outdated.

You have been able to use braces for years. And nodejs projects have better names? Yeah right.

This is about 15 years out of date.

Quality not quantity. There are so many bad npm libraries out there. Python is a trash fire of 2.7 stuff that won't even run anymore. Ruby is usually pretty good. There are good libraries for most major use cases.

Seems like you didn't do any basic research or read the article. There have been continuous optimizations in Ruby. Version 3.2 is quite fast and even has a JIT. Most of the rumors and misapprehensions of bad performance date to 1.9 era.

For my money, and I've been programming since the 90s and software engineering professionally since 2008, Ruby on Rails is a great solution for a company to build a form-driven web application. In many cases the performance is quite good and you can get something built that works well with a short development time. The community can't be beat and the libraries have a lot of rich functionality. Most of the myths and canards about Ruby and Rails are no longer true, if they even ever were. The main problem with Ruby is that it doesn't have a Google or a Microsoft behind it, although Github uses tons of Ruby. As does Shopify. As does Apple.

Of course they do. Propaganda has been helping, for example, the election of Marcos in the Philippines or the rise of fascism abroad. Nationalists and paid trolls brigade Wikipedia to try to push a narrative. Google has in many cases amplified misinformation by relying too heavily on so-called "machine learning" algorithms - that have a high error rate and do not learn as a human learns.