Ruby 3.0 Released With ~3x The Performance

[bofkentucky]

All the interpreted languages and their package managers end up suffering this "what version" problem when talking about LTS OS support. Perl (and cpan) suffered through it over the years, PHP. ruby had its struggles back in the day (everyone remember dueling 1.8 and 1.9 in ubuntu and debian), python with its drawn out migration from 2.x to 3.x. Its 2020 and just as an example, Datadog has to ship a canned python environment with their agent because you never know if python 2.x or 3.y is installed (properly), if pip is useable or you should load up OS provided python modules. I think most of the power of lambda on aws or cloud functions on google is getting the devs a useable environment without your sysadmins pulling their hair out getting the execution environment to match something usable by the developers.

[bug77]

Python is garbage. Last time college professors sold us on java now it's python. You can't trust college professors or phds.

Academic language are first and foremost... academic. Easy to grasp, easy to get something done. They're a starting point, not the end of the journey.
And while no language is without fault (especially if you insist on using it for the wrong job), both Java and Python are easy to learn and read. Both have extensive 3rd party libraries. Python also has bindings for pretty much whatever you can think of.

Go with nim, try distributing your python code/app on multiple OSes and they youll see the mess that is python(oh which version? oh what mess), with nim there is one file to copy, done, app distributed.
nim FTW.

I'm not sure whether you realize it, but the only advantage you listed for nim, is not specific to nim. It applies to any compiled language. It even applies to Java (if using GraalVM + native-image).

[polarathene]

If you're interested in a systems language you want to look into Rust which draws inspiration from Rust

..wut?

[onlyLinuxLuvUBack]

Because, for Python, the JIT is a separate project: PyPy.

...and PyPy is a little over 4 times faster than CPython. (Excepting, of course, execution patterns that don't give a JIT a chance to latch onto repeated code and translate it to machine code in time for it to help.)

Python is garbage. Last time college professors sold us on java now it's python. You can't trust college professors or phds.
Go with nim, try distributing your python code/app on multiple OSes and they youll see the mess that is python(oh which version? oh what mess), with nim there is one file to copy, done, app distributed.
nim FTW.

[bug77]

Just a note: If I was asked to choose the best programming language from the top 20 (https://www.tiobe.com/tiobe-index) I wouldn't choose any of them because all of them are missing features I consider essential for solving hard problems.

Makes sense. Because the languages on that list only got popular because of their inability to solve problems, right?

Or maybe you're just a manager and your language of choice must be able to handle "get this done by next week or start looking for another job".

[uid313]

While that may all be true (libraries and frameworks matter to the ecosystem of a language (what would python be with no libraries, or rust without any crates?)), that tends to discount that some large organizations, such as github, still run their core on ruby.

Yeah, GitHub is stuck with their old Ruby code base, like Facebook is stuck with their PHP heritage. But Ruby has no future and no newcomers, and is only used for Ruby on Rails and old legacy code bases that went into production and now still runs that.

[JackLilhammers]

It is a strange post he made: his list goes from assembler to C to some OO languages to scripting languages to Matlab. If super low level is not good enough for him and super high level is not good enough for him and in-the-middle is not good enough for him then I have to echo your question.

Perhaps he can only solve problems with functional languages (ie: let the runtime figure out his DWIM)?

Well, it's not "his" list, the TIOBE Index is a popularity ranking based on search engines data.
That's why languages so different are ranked together.
However it doesn't mean much more than that. Niche languages or legacy ones get little consideration, but they can be absolutely fundamental to the industry where are used

[hoohoo]

What are the features you're missing?
The only things I can think of are reliability guarantees and maybe some purely functional stuff.
However I'd argue that your definition of essential might be very personal

It is a strange post he made: his list goes from assembler to C to some OO languages to scripting languages to Matlab. If super low level is not good enough for him and super high level is not good enough for him and in-the-middle is not good enough for him then I have to echo your question.

Perhaps he can only solve problems with functional languages (ie: let the runtime figure out his DWIM)?

[hoohoo]

Just a note: If I was asked to choose the best programming language from the top 20 (https://www.tiobe.com/tiobe-index) I wouldn't choose any of them because all of them are missing features I consider essential for solving hard problems.

I must say, I'm not entirely sure what to make of a language ranking that pits procedural/OO languages (C, C++, PHP, Python,...) against data definition & data manipulation languages (SQL, PL/SQL), and then against super-high-level scientist math toolkits (Matlab).

The list is perhaps not very useful.

[ssokolow]

Meanwhile Python performance stuck

Because, for Python, the JIT is a separate project: PyPy.

...and PyPy is a little over 4 times faster than CPython. (Excepting, of course, execution patterns that don't give a JIT a chance to latch onto repeated code and translate it to machine code in time for it to help.)