Perhaps this is just an “old man yells at cloud” moment...
Fellow "old man" reporting for duty.
I see a similar set of problems with both Python and JavaScript today (and with various other languages as well, but IME Python and JS are the worst offenders). Both have proven to be useful for various applications. Each started with quite a distinctive style. There were plenty of warts in each case, but overall the "feel" of each language was well understood, and there was a certain simplicity about them that made them accessible to new programmers or to programmers who primarily used other languages. However, over time, people have applied the kitchen sink approach to both language and tooling.
Functional programming is all the rage, and these languages have lambdas (albeit very clumsy ones), so let's all write lots of FP-style code in them! Except you can't, because they are horrible for writing FP-style code, because they lack FP features, they lack efficient and clear syntax for FP idioms, and they do not provide some of the guarantees that benefit FP.
Similarly, a lot of programmers are realising that static typing has its merits when you're trying to build serious software at scale, so let's all add type annotations and start marking up our code! Except that both the existing type systems and idiomatic use of these languages are built around dynamic behaviour, so they lack powerful tools for constructing new types, and they lack efficient and clear syntax for specifying static types, and they lack comprehensive type inference, and the continued presence of dynamic typing means you can't prove a lot of otherwise useful properties of your code anyway.
The amount of tooling expected to do even basic development with these languages today is absurd. Professional-quality programming ought to need no more than a compiler/interpreter, a decent editor and a decent source control system to be productive. On the next level, we might find a tool to automate running tests and reporting the results, a debugger and a profiler; I count these as secondary because it should really be possible to work reasonably well without them but they can make common tasks significantly more efficient and/or accurate. The lack of a package manager tool in both of these lists is not an accident.
I think just as much of today's programming industry has forgotten about the benefits of stability, backward and forward compatibility, and standardisation in the software we make, we have also forgotten about them in our own tools. Do we really need 27 new frameworks for building web apps this year? Is having 41 libraries to do basic unit testing that each use slightly different syntax and vary in <5% of the features they provide really worth the added complexity in the ecosystem? How on earth does JS not have a decent standard library yet?
Maybe it's time to slow down, and make carefully considered choices about the tools and libraries we use in our work, instead of instinctively reaching for that package manager and installing... well, let's be honest, most of the time we don't really know exactly what we're installing, do we?
Fellow "old man" reporting for duty.
I see a similar set of problems with both Python and JavaScript today (and with various other languages as well, but IME Python and JS are the worst offenders). Both have proven to be useful for various applications. Each started with quite a distinctive style. There were plenty of warts in each case, but overall the "feel" of each language was well understood, and there was a certain simplicity about them that made them accessible to new programmers or to programmers who primarily used other languages. However, over time, people have applied the kitchen sink approach to both language and tooling.
Functional programming is all the rage, and these languages have lambdas (albeit very clumsy ones), so let's all write lots of FP-style code in them! Except you can't, because they are horrible for writing FP-style code, because they lack FP features, they lack efficient and clear syntax for FP idioms, and they do not provide some of the guarantees that benefit FP.
Similarly, a lot of programmers are realising that static typing has its merits when you're trying to build serious software at scale, so let's all add type annotations and start marking up our code! Except that both the existing type systems and idiomatic use of these languages are built around dynamic behaviour, so they lack powerful tools for constructing new types, and they lack efficient and clear syntax for specifying static types, and they lack comprehensive type inference, and the continued presence of dynamic typing means you can't prove a lot of otherwise useful properties of your code anyway.
The amount of tooling expected to do even basic development with these languages today is absurd. Professional-quality programming ought to need no more than a compiler/interpreter, a decent editor and a decent source control system to be productive. On the next level, we might find a tool to automate running tests and reporting the results, a debugger and a profiler; I count these as secondary because it should really be possible to work reasonably well without them but they can make common tasks significantly more efficient and/or accurate. The lack of a package manager tool in both of these lists is not an accident.
I think just as much of today's programming industry has forgotten about the benefits of stability, backward and forward compatibility, and standardisation in the software we make, we have also forgotten about them in our own tools. Do we really need 27 new frameworks for building web apps this year? Is having 41 libraries to do basic unit testing that each use slightly different syntax and vary in <5% of the features they provide really worth the added complexity in the ecosystem? How on earth does JS not have a decent standard library yet?
Maybe it's time to slow down, and make carefully considered choices about the tools and libraries we use in our work, instead of instinctively reaching for that package manager and installing... well, let's be honest, most of the time we don't really know exactly what we're installing, do we?