It's "outstanding features" were creating a syntax that a lot of people can quickly grok (not just programmers) and the way they allowed for extensions to be written in C.
I believe it also popularized the concept of "there should be one obvious way of doing things", in stark contrast with Perl; "there's more than one way to skin a cat".
It's easy to look at this now and say that it's nothing interesting, because it already pushed the boundaries.
All that being said, Python's age is irrelevant when discussing the future of programming.
It's one of the principles from the Zen of Python: https://en.wikipedia.org/wiki/Zen_of_Python#Principles
I agree with your comment, all of these taken together were a medium-sized revolution at the time. It's not for nothing that Python is the most popular language for teaching programming nowadays.
This. We tend to underestimate what it takes to create something easy to use.
Features don’t make a language good or important. Its the invisible broth the features live in that make them useful. The cohesive nature of python is something few languages can touch. The features in python though not always innovative by themselves work together like magic.
Most people stuck in the initial hype phase of FP often miss these things, but python is special & shouldn’t be discarded because its lack of hyped features of the day.
I am not sure I follow that at all. I have not known Python to be cohesive, like at all. Just look at how to do async/concurrency in Python. (a) you can't really, and (b) there are several ways to do with all sorts of quirks.
> Most people stuck in the initial hype phase of FP often miss these things, but python is special & shouldn’t be discarded because its lack of hyped features of the day.
People saying this usually just equate functional programming with Haskell. I know van Rossum does because he has said so himself. Functional-first programming, on the other hand, is a totally different ballgame. There isn't a hype phase there because such languages, like F# sit, in about the sweetest spot you can for programming languages. Also, the features Python lacks were around for at least a decade before it. It's not like they're just now the soup of the day.
This sums it up very nicely, I think. Without switching a language I can automate deployments, do machine learning tasks, implement ideas quickly, create backends, interface with most C libs easily and query apis.
Personally, as long as I have a good linter, I prefer JavaScript for general purpose programming but there are some areas where other languages just dominate.
I don't blame Python for such issues, i know it's human error. But why no such issues happen in other languages ?
More beginners into programming is a good thing. But the worst things happen when they don't know a better way to do things. Because "There's only one Python way to do things".
That's the reason i avoid Python at all cost to teach programming to newscomers.
You can look at the topic-shortcuts on the video timeline to see if any of these topics are of interest to you.
You obviously haven’t looked through code bases in other languages if you never seen these issues outside of python. Personally I have observed them in C, C++, Java, Haskell, R, Python, PHP, JavaScript , Matlab, Maple, Mathematica, Rust, heck even in lab view. And the same type of problems even translate to poorly developed excel sheets.
The only plausible explanation for you never having seen any such issues outside of python is that you have never looked.
You can create dumb code in any language.
- First class functions (pretty much all languages)
- First class null values through Option/Maybe (several languages use nullable types like Typescript but it's an approximation to the real thing)
- Unit type instead of hacks like 'void' (Kotlin, Rust)
- Expressions over statements (ex. if..then..else is an expression, and you don't rely on hacks like ternary expressions) (Kotlin, Rust, Scala)
- Statically typed with full type inference (Kotlin, Rust, Scala)
- Pattern matching (Java and C# are continuously extending this feature)
- Green threads as default (Go, soon Java)
- Decouple data from interfaces through Type Classes/Traits (Rust, C#'s static virtual members in interfaces is a start)
- Generic programming. Do not confuse this with parametric polymorphism (I think Rust has an approximation through macros)
- Effects (like IO) as first class values (Scala's ecosystem tries to mimic this)
"Readability counts" might be the one exception, at least early on in the language's lifetime. Nowadays it could be argued that even that is diminishing.
But readability is often overlooked by most programming languages, which is a mistake. Programmers (should) spend more time reading than writing code, so it's very important. It's also very difficult to get right, being so subjective, but GvR has a good eye for it, and he made good design choices to balance it with expressiveness.
Readability is also crucial for a language being welcoming to newcomers, and new programmers in general. I think that Python is still the best first language for people learning to program. This is likely the biggest reason for its adoption and popularity today.
Especially when the idea of there being one obvious way of doing things is contradicted in probably the most used package of pandas, and even built in things like list comprehensions vs loops vs itertools, then couple that with the way it does variable passing.
Can't really agree that it is easier to read for laymen than other c-like languages of its day either, with the use of underscores for class methods and indentation defining scope.
Honestly if I wanted to really know why python was successful interviews with the authors of the notebook, pandas and the AI libraries would be my targets, cause in my view they're the ones that made the language a success.
Even then, Elixir notebooks with Livebook and F# notebooks with .NET Interactive (which are actually polyglot notebooks with several languages that can share data) are ahead of plain old Python notebooks.