I code both typescript and python professionally, and python is almost as much of a mess, just a different kind of mess. The package manager ecosystem is all over the place, nobody is agreeing on a build system, and the type system is still unable to represent fairly simple concepts when it comes to function typing. Also tons of libraries just ignore types altogether. I love it, but as a competitor to JavaScript in the messiness department it’s not a good horse.
I’m going to code in Python for my job soon, do you have links on how TF I should manage dependencies? I can’t stand the bloat of virtual environments or Anaconda sorry.
I need to make myself a stack python but I don’t want to look at this madness. At least Typescript benefits from a high quality ecosystem of frontend tooling with Vite/React/ESLint/ESBuild and it’s getting even better with the new runtimes features :)
I need to learn about:
built-in tools provided by the language (practice on the job and the power of Google & LLMs will help)
ecosystem
good practices
Why is their a build system for an interpreted language ? Do you have bundling concerns like in JS or you somehow compile Python code now?
You need to get over the bloat of virtual environments. It’s the same as node_modules and it’s completely necessary if you want more than a single python project to live on your machine.
I personally use poetry as my dependency manager and build tool. It’s not perfect but it’s a lot better than pipenv or just rawdogging pip like a maniac.
uv is the new hotness, but I haven’t tried it so can’t vouch. People seem to like it though.
JavaScript is also an interpreted language with tons of build tools. The reason to have one for python is mainly about packaging and code distribution, so same as JavaScript. If you want to distribute a program you probably don’t want to just point people to a GitHub repo, and if you want to publish a package on pypi it needs to be bundled correctly.
For ecosystem there isn’t much I can do for you, it completely depends on what you’ll be working on. Baseline you want pydantic for parsing objects, assuming some APIs will be involved. You want black for code formatting, flake8 for linting, pytest for testing. If you’re gonna write your own APIs you can’t go wrong with fastapi, which works great with pydantic. For nice console stuff there’s click for building cli apps and rich and textual for console output and live console apps respectively.
People are actively trying to replace flake8 and black with feature compatible stuff written in rust but again I haven’t tried those so can’t vouch.
Coming from react you’re gonna need to pretty quickly switch gears to thinking more object oriented. You’re gonna be annoyed at how you can’t just quickly declare a deeply nested interface, that’s just how it is. The biggest change other than object oriented thinking will probably be decorators. Typescript had them experimentally and only for classes, python has them for classes and functions natively. They’re a bit tricky to wrap your mind around when you want to write your own, but not too bad. A lot of Google hits will be outdated on this front. Google specifically “decorators ParamSpec” to see how to make them properly.
All documentation is optional and ignored at runtime, that doesn’t mean you shouldn’t do it. If your library doesn’t have type hints I’m just not gonna use it, I don’t have the time to figure out what you accept or return.
It means they have the option to do it or not do it and you have the option to not use it, which clearly your exercising. I’ve personally never had a situation where a lack of type hints slowed me down even a little.
I dunno if you’re being deliberately obtuse, but just in case you really did miss his point: the fact that type hints are optional (and not especially popular) means many libraries don’t have them. It’s much more painful to use a library without type hints because you lose all of their many benefits.
This obviously isn’t a problem in languages that require static types (Go, Rust, Java, etc…) and it isn’t a problem with Typescript because static types are far more popular in JavaScript/Typescript land so it’s fairly rare to run into a library that doesn’t have them.
And yeah you can just not use the library at all but that’s just ignoring the problem.
True, but if you’re looking at a Python library that doesn’t have type hints in 2024, then chances are that it’s not very good and/or not very well maintained.
My biggest pet peeve is the complete inability to annotate a set of known exceptions that a function raises in a machine readable way. The discussion about it is quite heated.
In fairness that approach hasn’t really worked in other languages. It was so unpopular in C++ that they actually removed the feature, which is almost unheard of. Java supports it too but it’s pretty rarely used in my experience. The only place I’ve seen it used is in Android. It’s unpopular enough there that Kotlin doesn’t support it.
Interesting. I’ve never felt a need for this, and as the other reply here said it was really unpopular in other languages.
I would have guessed you would have said something about how it’s annoying to type callable arguments, and how Protocol exists but doesn’t seem that widely known.
Definitely those used to be pain points, but they do exist now so type erasure after decorator application isn’t a problem anymore, which used to be another huge one for me.
The discussion around how unpopular it was in other languages seems like such an obvious side track to me. Typing in general went out of fashion and then made a comeback when it was opt-in, why wouldn’t the same apply to exceptions? Of course I’m not wanting warnings in every func call because of a potential MemoryCorruptionError, but if a library has some set of known exceptions as a de facto part of its interface then that’s currently completely unknown to me and my static type checker.
One kinda bad example is playwright. Almost all playwright functions have the chance to raise a TimeoutError, but even if you know this you’ll probably shoot yourself in the foot at least once because it’s not the built-in TimeoutError, oh no, it’s a custom implementation from the library. If you try to simply try...except TimeoutError:, the exception will blow right by you and crash your script, you’ve got to import the correct TimeoutError.
If it was properly typed then pyright would be able to warn you that you still need to catch the other kind of TimeoutError.
It’s a bad example because like I said almost all playwright functions can raise this error so you’d get a lot of warnings, but it also demonstrates well the hidden interface problem we have right now, and it’s the most recent one that screwed me, so it’s the one I remember off the top of my head.
Yeah, they’re useful when developing, which is why it’s so frustrating when libraries don’t implement types. I’m developing and I’m trying to use a tool that supposedly fits a use case I have, but the tool didn’t come with instructions so it’s practically useless to me. I could open the tool up and look at its guts to figure it out but are you kidding me no, I’m not going back to the stone age for your tool.
Typescript is far nicer than Python though. Well I will give Python one point: arbitrary precision integers was absolutely the right decision. Dealing with u64s in Typescript is a right pain.
But apart from that it’s difficult to see a single point on which Python is clearly better than Typescript:
Static typing. Pyright is great but it’s entirely optional and rarely used. Typescript obviously wins here.
Tooling. Deno is fantastic but even if we regress to Node/NPM it’s still a million miles better than the absolute dog shit pile of vomit that is Pip & venv. Sorry Python but admit your flaws. uv is a shining beacon of light here but I have little hope that the upstream Python devs will recognise that they need to immediately ditch pip in favour of officially endorsing uv. No. They’ll keep it on the sidelines until the uv devs run out of hope and money and give up.
Performance. Well I don’t need to say more.
Language sanity. They’re pretty on par here I think - both so-so. JavaScript has big warts (the whole prototype system was clearly a dumb idea) but you can easily avoid them, especially with ESLint. But Python has equally but warts that Pylint will tell you about, e.g. having to tediously specify the encoding for every file access.
Libraries & ecosystem. Again I would say there’s no much in it. You’d obviously be insane to use Python for anything web related (unless it’s for Django which is admittedly decent). On the other hand Python clearly dominates in AI, at least if you don’t care about actually deploying anything.
What is so bad about virtual environments? I found them to be really nice and useful when I developed in Python over about 5-ish years. It was really nice being able to have separate clean environments for installing libraries and executing things.
Granted, I only used Python as a solo developer, so if there are shortcomings that emerge when working with other developers, then I would not be aware of them…
Edit: also, performance is a bit more of a subtle topic. For numerical logic, Python actually is (probably) much better than a lot of its competitors thanks to numpy and numexpr. For conditional logic, I would agree that it’s not the best, but when you consider developer velocity, it’s a clearly worthwhile tradeoff since frameworks like Django are so popular.
They’re a solution to a self-inflicted problem. They’re only “really nice and useful” if you accept that having your projects stomp all over each others’ libraries and environments is normal.
If projects were self-contained from the outset then you wouldn’t need an additional tool to make them so.
I write mostly Python for 5 years and uv is indeed the best thing that happened to the Python landscape during this period.
I disagree that typescript is far nicer; even syntax-wise, type annotated Python seems much easier to read, write, and refactor; but I’ll give that Python needs to ditch pip and “requirements.txt” for good.
the absolute dog shit pile of vomit that is Pip & venv
I’ve worked professionally in python for several years and I don’t think it’s ever caused a serious problem. Everything’s in docker so you don’t even use venv.
Operator '+' cannot be applied to types 'number[]' and 'number[]'.
We’re talking about Typescript here. Also I did say that it has some big warts, but you can mostly avoid them with ESLint (and Typescript of course).
Let’s not pretend Python doesn’t have similar warts:
>>>x = -5>>>y = -5>>>x is y
True
>>>x = -6>>>y = -6>>>x is y
False
>>>x = -6; y = -6; x is y
True
>>>isinstance(False, int)
True
>>>[f() for f in [lambda: i for i inrange(10)]]
[9, 9, 9, 9, 9, 9, 9, 9, 9, 9]
There’s a whole very long list here. Don’t get be wrong, Python does a decent job of not being crazy. But so does Typescript+ESLint.
I’ve worked professionally in python for several years and I don’t think it’s ever caused a serious problem. Everything’s in docker so you don’t even use venv.
“It’s so bad I have resorted to using Docker whenever I use Python.”
No because I am not using Python to make a web app. That’s not the only thing people write you know…
JavaScript is so bad you’ve resorted to using a whole other language: Typescript
Well yeah. Typescript isn’t really a new language. It’s just type annotations for JavaScript (except for enums; long story). But yes JavaScript is pretty bad without Typescript.
But Typescript isn’t a cop-out like Docker is.
But the language it’s built on top of it is extremely warty. Maybe we agree on that.
Yeah definitely. You need to ban the warts but Typescript & ESLint do a pretty good job of that.
I mean I would still much rather write Dart or Rust but if I had to pick between Typescript and Python there’s absolutely no way I’d pick Python (unless it was for AI).
Why would you add two arrays like that?
Because I want to combine two lists.
The is operator is for identity, not equality. Your example is just using it weirdly in a way that most people wouldn’t do.
No because I am not using Python to make a web app. That’s not the only thing people write you know…
Most of what I’ve worked on has been webapps or services that support them :shrug:
Typescript and Python there’s absolutely no way I’d pick Python (unless it was for AI).
Agree to disagree then. We could argue all day but I think it’s mostly opinion about what warts and tradeoffs are worth it, and you don’t seem like you have no idea what you’re talking about. Sometimes I meet junior developers who have only ever used javascript, and it’s like (to borrow another contentious nerd topic) like meeting someone who’s only ever played D&D talking about game design.
The is operator is for identity, not equality. Your example is just using it weirdly in a way that most people wouldn’t do.
The + operator is for numbers or strings, not arrays. Your example is just using it weirdly in a way that most people wouldn’t do.
I’m not defending Javascript’s obviously terrible behaviour there. Just pointing out that Python has obviously terrible behaviours too. In both cases the solution is “don’t do that, and use static analysis to make sure you don’t do it accidentally”.
Sometimes I meet junior developers who have only ever used javascript, and it’s like (to borrow another contentious nerd topic) like meeting someone who’s only ever played D&D talking about game design.
Yeah I think you can generalise that to “have only ever used one language”. I would say Python and Javascript are pretty close on the “noob level”. By which I mean if you meet someone who has only ever written C++, Java, or Rust or whatever they’re going to be a class above someone who has only ever written Python or Javascript.
Scala does look nice. Just a quick syntax view makes me want to give it a whirl when I want an alternative to Python. I used to code in C++, and C#. I use G’MIC (DSL) as my main. Scala seems right up my alley.
Thank god, Javascript is a mess.
I’ll still plug Scala for having the beauty of Python, the ecosystem of Java, the correctness of Rust, the concurrency of Go, and the power of Lisp.
I code both typescript and python professionally, and python is almost as much of a mess, just a different kind of mess. The package manager ecosystem is all over the place, nobody is agreeing on a build system, and the type system is still unable to represent fairly simple concepts when it comes to function typing. Also tons of libraries just ignore types altogether. I love it, but as a competitor to JavaScript in the messiness department it’s not a good horse.
I’m going to code in Python for my job soon, do you have links on how TF I should manage dependencies? I can’t stand the bloat of virtual environments or Anaconda sorry.
I need to make myself a stack python but I don’t want to look at this madness. At least Typescript benefits from a high quality ecosystem of frontend tooling with Vite/React/ESLint/ESBuild and it’s getting even better with the new runtimes features :)
I need to learn about:
Why is their a build system for an interpreted language ? Do you have bundling concerns like in JS or you somehow compile Python code now?
You need to get over the bloat of virtual environments. It’s the same as node_modules and it’s completely necessary if you want more than a single python project to live on your machine.
I personally use poetry as my dependency manager and build tool. It’s not perfect but it’s a lot better than pipenv or just rawdogging pip like a maniac. uv is the new hotness, but I haven’t tried it so can’t vouch. People seem to like it though.
JavaScript is also an interpreted language with tons of build tools. The reason to have one for python is mainly about packaging and code distribution, so same as JavaScript. If you want to distribute a program you probably don’t want to just point people to a GitHub repo, and if you want to publish a package on pypi it needs to be bundled correctly.
For ecosystem there isn’t much I can do for you, it completely depends on what you’ll be working on. Baseline you want
pydanticfor parsing objects, assuming some APIs will be involved. You wantblackfor code formatting,flake8for linting,pytestfor testing. If you’re gonna write your own APIs you can’t go wrong withfastapi, which works great with pydantic. For nice console stuff there’sclickfor building cli apps andrichandtextualfor console output and live console apps respectively.People are actively trying to replace flake8 and black with feature compatible stuff written in rust but again I haven’t tried those so can’t vouch.
Coming from react you’re gonna need to pretty quickly switch gears to thinking more object oriented. You’re gonna be annoyed at how you can’t just quickly declare a deeply nested interface, that’s just how it is. The biggest change other than object oriented thinking will probably be decorators. Typescript had them experimentally and only for classes, python has them for classes and functions natively. They’re a bit tricky to wrap your mind around when you want to write your own, but not too bad. A lot of Google hits will be outdated on this front. Google specifically “decorators ParamSpec” to see how to make them properly.
Good luck in your new job, you’ll be grand!
They ignore types all together because typing is optional in python.
All documentation is optional and ignored at runtime, that doesn’t mean you shouldn’t do it. If your library doesn’t have type hints I’m just not gonna use it, I don’t have the time to figure out what you accept or return.
It means they have the option to do it or not do it and you have the option to not use it, which clearly your exercising. I’ve personally never had a situation where a lack of type hints slowed me down even a little.
I dunno if you’re being deliberately obtuse, but just in case you really did miss his point: the fact that type hints are optional (and not especially popular) means many libraries don’t have them. It’s much more painful to use a library without type hints because you lose all of their many benefits.
This obviously isn’t a problem in languages that require static types (Go, Rust, Java, etc…) and it isn’t a problem with Typescript because static types are far more popular in JavaScript/Typescript land so it’s fairly rare to run into a library that doesn’t have them.
And yeah you can just not use the library at all but that’s just ignoring the problem.
True, but if you’re looking at a Python library that doesn’t have type hints in 2024, then chances are that it’s not very good and/or not very well maintained.
Well, indeed. Unfortunately there are still a fair number of them. The situation is definitely improving at least.
what do you mean by this?
My biggest pet peeve is the complete inability to annotate a set of known exceptions that a function raises in a machine readable way. The discussion about it is quite heated.
In fairness that approach hasn’t really worked in other languages. It was so unpopular in C++ that they actually removed the feature, which is almost unheard of. Java supports it too but it’s pretty rarely used in my experience. The only place I’ve seen it used is in Android. It’s unpopular enough there that Kotlin doesn’t support it.
Interesting. I’ve never felt a need for this, and as the other reply here said it was really unpopular in other languages.
I would have guessed you would have said something about how it’s annoying to type callable arguments, and how
Protocolexists but doesn’t seem that widely known.Definitely those used to be pain points, but they do exist now so type erasure after decorator application isn’t a problem anymore, which used to be another huge one for me.
The discussion around how unpopular it was in other languages seems like such an obvious side track to me. Typing in general went out of fashion and then made a comeback when it was opt-in, why wouldn’t the same apply to exceptions? Of course I’m not wanting warnings in every func call because of a potential MemoryCorruptionError, but if a library has some set of known exceptions as a de facto part of its interface then that’s currently completely unknown to me and my static type checker.
One kinda bad example is playwright. Almost all playwright functions have the chance to raise a TimeoutError, but even if you know this you’ll probably shoot yourself in the foot at least once because it’s not the built-in TimeoutError, oh no, it’s a custom implementation from the library. If you try to simply
try...except TimeoutError:, the exception will blow right by you and crash your script, you’ve got to import the correct TimeoutError. If it was properly typed then pyright would be able to warn you that you still need to catch the other kind of TimeoutError. It’s a bad example because like I said almost all playwright functions can raise this error so you’d get a lot of warnings, but it also demonstrates well the hidden interface problem we have right now, and it’s the most recent one that screwed me, so it’s the one I remember off the top of my head.types are always ignored at runtime, they’re only useful when developing
Yeah, they’re useful when developing, which is why it’s so frustrating when libraries don’t implement types. I’m developing and I’m trying to use a tool that supposedly fits a use case I have, but the tool didn’t come with instructions so it’s practically useless to me. I could open the tool up and look at its guts to figure it out but are you kidding me no, I’m not going back to the stone age for your tool.
Pydantic offers awesome runtime validation (using Rust).
deleted by creator
Typescript is far nicer than Python though. Well I will give Python one point: arbitrary precision integers was absolutely the right decision. Dealing with u64s in Typescript is a right pain.
But apart from that it’s difficult to see a single point on which Python is clearly better than Typescript:
uvis a shining beacon of light here but I have little hope that the upstream Python devs will recognise that they need to immediately ditch pip in favour of officially endorsinguv. No. They’ll keep it on the sidelines until theuvdevs run out of hope and money and give up.What is so bad about virtual environments? I found them to be really nice and useful when I developed in Python over about 5-ish years. It was really nice being able to have separate clean environments for installing libraries and executing things.
Granted, I only used Python as a solo developer, so if there are shortcomings that emerge when working with other developers, then I would not be aware of them…
Edit: also, performance is a bit more of a subtle topic. For numerical logic, Python actually is (probably) much better than a lot of its competitors thanks to numpy and numexpr. For conditional logic, I would agree that it’s not the best, but when you consider developer velocity, it’s a clearly worthwhile tradeoff since frameworks like Django are so popular.
They’re a solution to a self-inflicted problem. They’re only “really nice and useful” if you accept that having your projects stomp all over each others’ libraries and environments is normal.
If projects were self-contained from the outset then you wouldn’t need an additional tool to make them so.
If I need to keep my Python environment separate I’d rather spin up a docker container. They make virtual environments pointless
But then you need to connect your IDE to the docker container. Doable, but often a PITA IME
A single extension and 1-2 clicks isn’t that much to me 🤷 I’ve been doing it painfree for a few years now
Maybe it’s gotten better in recent times…it was always disconnecting and needing to be restarted
I write mostly Python for 5 years and uv is indeed the best thing that happened to the Python landscape during this period.
I disagree that typescript is far nicer; even syntax-wise, type annotated Python seems much easier to read, write, and refactor; but I’ll give that Python needs to ditch pip and “requirements.txt” for good.
[1] + [2] "12"A sane language, you say.
const foo = 'hello' const bar = { foo: 'world'} console.log(bar) // { "foo": "world" }I’ve worked professionally in python for several years and I don’t think it’s ever caused a serious problem. Everything’s in docker so you don’t even use venv.
Yes:
Operator '+' cannot be applied to types 'number[]' and 'number[]'.We’re talking about Typescript here. Also I did say that it has some big warts, but you can mostly avoid them with ESLint (and Typescript of course).
Let’s not pretend Python doesn’t have similar warts:
>>> x = -5 >>> y = -5 >>> x is y True >>> x = -6 >>> y = -6 >>> x is y False >>> x = -6; y = -6; x is y True>>> isinstance(False, int) True>>> [f() for f in [lambda: i for i in range(10)]] [9, 9, 9, 9, 9, 9, 9, 9, 9, 9]There’s a whole very long list here. Don’t get be wrong, Python does a decent job of not being crazy. But so does Typescript+ESLint.
“It’s so bad I have resorted to using Docker whenever I use Python.”
Why would you use the
isoperator like that?The lambda thing is from late binding, which I’ve had come up at work once. https://docs.python-guide.org/writing/gotchas/#late-binding-closures.
Do you not use containers when you deploy ? Everywhere I’ve worked in the past like 10 years has moved to containers.
Also this is the same energy as “JavaScript is so bad you’ve resorted to using a whole other language: Typescript”
To your point, typescript does solve a lot of problems. But the language it’s built on top of it is extremely warty. Maybe we agree on that.
Why would you add two arrays like that?
No because I am not using Python to make a web app. That’s not the only thing people write you know…
Well yeah. Typescript isn’t really a new language. It’s just type annotations for JavaScript (except for enums; long story). But yes JavaScript is pretty bad without Typescript.
But Typescript isn’t a cop-out like Docker is.
Yeah definitely. You need to ban the warts but Typescript & ESLint do a pretty good job of that.
I mean I would still much rather write Dart or Rust but if I had to pick between Typescript and Python there’s absolutely no way I’d pick Python (unless it was for AI).
The
isoperator is for identity, not equality. Your example is just using it weirdly in a way that most people wouldn’t do.Agree to disagree then. We could argue all day but I think it’s mostly opinion about what warts and tradeoffs are worth it, and you don’t seem like you have no idea what you’re talking about. Sometimes I meet junior developers who have only ever used javascript, and it’s like (to borrow another contentious nerd topic) like meeting someone who’s only ever played D&D talking about game design.
The + operator is for numbers or strings, not arrays. Your example is just using it weirdly in a way that most people wouldn’t do.
I’m not defending Javascript’s obviously terrible behaviour there. Just pointing out that Python has obviously terrible behaviours too. In both cases the solution is “don’t do that, and use static analysis to make sure you don’t do it accidentally”.
Yeah I think you can generalise that to “have only ever used one language”. I would say Python and Javascript are pretty close on the “noob level”. By which I mean if you meet someone who has only ever written C++, Java, or Rust or whatever they’re going to be a class above someone who has only ever written Python or Javascript.
Scala does look nice. Just a quick syntax view makes me want to give it a whirl when I want an alternative to Python. I used to code in C++, and C#. I use G’MIC (DSL) as my main. Scala seems right up my alley.
The only problem is that it also has the complexity of C++.
Not the foot-shooting complexity though, just the extra-power complexity
You can enable the foot-shooting complexity by writing modules for Python in C++, since it’s very easy to do. Why do I know this? Well…