I have to disagree with that.

Don't tell that to them: https://www.openstack.org/

Or them:

https://numpy.org/
https://pandas.pydata.org/

I think Python is great for A LOT of serious software. You just need to be mindful of the performance impact and the interpreter/package dependencies.

Python is actually a really nice language for writing code in.

Some users find Python's dynamic typing to be annoying for larger code bases, but at least type mismatches tend to result in an error.

Explanation:

* The following code works thanks to dynamic typing
x = 3
x = "three"
(In other words, a variable can change type. Note, also, that variables don't need to be declared ahead of time, which cau cause issues if there's a typo.)

* You can't, however do the following:
x = 1 + "1"
(Perl silently converts "1" to a number and sets x to 2. JavaScript silently converts 1 to a string, and sets x to "11". Python will give an error to the effect of "can't add a number and a strong".)

(It's better to get an error message right away than have the program silently run and have a bug you need to track down later.) Actually, I believe Python grew up on Linux or some UNIX variant. In fact, Python is in the standard install of many Linux distributions, and is even essential in some. For example. Gentoo's emerge program, which you use to install software, is written in Python (which means that removing Python, which emerge will refuse to do, would prevent you from easily re-installing it).

Also, I believe the language is actually named after Monty Python. There's a trade-off between programmer time and CPU time. In many cases, programmer time is more important, and Python takes less of it than some other languages. Hence, if you want to get the project completed faster, or if you just want to make a prototype before you re-write in a higher performance language, Python is the ideal choice (or at least close to ideal).

Java is also opinionated about your choice of programming paradigm.

In particular, Java essentially forces object oriented programming. You can't even write a basic Hello World program without declaring a class and a member function, with something like 8 keywords that someone unfamiliar with OOP is not going to understand. By contrast, in C, all you need to do is write a single function, with void parameters and a simple return type, and it will run.

(C is a decent language, at least until you're dealing with strings or dynamic memory.)

You don't need any for simple scripts; it's only when you scale thing up that you'll find the need for that sort of thing.

In particular, simple shell scripts and Python scripts don't need such tools, particularly if the project is small enough to keep it in a single file while still having it be manageable.

There is a tradeoff, but the tradeoff is not binary, its a scale.

In terms of cpu/dev time tradeoff:

C > Golang > NodeJS > Python

It will take me about 3 times longer to write the same functionality in C than Python, for like 100x the performance.

It might take me 1.3-1.5 times longer to write the same functionality in Golang, for maybe 30x the performance.

For a lot of use-cases, that tradeoff favors Golang, but not C, over Python.

Also, there is tooling to consider.

If I want to write a Terraform provider, there is an sdk in Golang, but not Python (that I'm aware of, if there is, its unofficial). That makes Golang A LOT faster to develop in to write a Terraform provider.

Also, if I want to run a client on user-machines, with Python, I need to bundle the interpreter and packages in the binary if I want a self-contained binary, which will bloat the size of the binary considerably (and is also an overhead in terms of packaging the final deliverable).

Then why is the Linux Kernel starting to get converted from C to Rust? (Read here.)

The kernel is different, plus the points I made aren't entirely bad. I also note that they're not re-writing the entire kernel in Rust; the first bits of Rust are going to be in drivers and similar leaf modules, not in the core that many things depend on.

Rust is a lot harder to write:
* For a kernel, this is probably OK, particularly since kernel code can be difficult to get write.

Harder for Rust code to compile:
* That's because Rust has safeguards to prevent certain types of bugs. Many of these bugs would likely cause a segmentation fault in userspace, but in the kernel, you don't have such nice behavior (though I might expect the kernel to trap the memory access and panic or oops if this happens).

Rust's stdlib being more bare bones than C's:
* stdlib is for user-space applications; we don't have C's stdlib available when working on the kernel, so this point doesn't really make sense at this point. You might have some functions like memcpy() available, but certainly none of the I/O or memory allocation functions (instead, you need to use in-kernel functions like kprintf() or kmalloc(), or access the hardware directly if you're a driver for that hardware). Similarly, you're not going to have rust's std, and will have to just use core, perhaps with some of the in-kernel functions adapted to Rust usage. (One nice thing about Rust is that it's easy to call C from Rust and vice versa, provided the Rust function has a C compatible signature.)

Rust code is more likely to be bug-free if it does compile:
* Isn't this a *good* thing, and a reason in favor of using Rust?

Wow, wow, wow, wow...

Hi dtgreene,

This statement really caught me. I've never heard about trade offs between programmers and CPU times.
Would you mind to provide some references about it please? Because I find it a total non-sense... but you know... my ignorance is the worst adviser, so I'd really like to read and learn about that to forge an opinion.

Thanks in advance

why would anybody convert a kernel that makes no sense at all

why not just build a new one better with modern design ?

Logically you can use just about any tool to make a website, at least at the simplest way. If i could feed output of say a awk program out to a port that is being sent to other people, and it happens to have html text, then you get the typical website results.

Getting input from what the Post/GET types are to make it actually usable would merely be a simple library or a way to accept said information. PHP you'd get that information in a variable as an array of data.

After that what else you use for database (or files) is mostly up to you. But that's a bare-bones minimal approach.

CPU time is much cheaper than developer time ;)
This is the main reason we target "good enough" code in enterprises, but hobbyist open-source software tend to have much higher quality standards.

---

You’re the only one here talking about the NT kernel…

talk when you reached 2% market share :)
that means never :P

Its simple. I need to deliver to production by January 12th and I cost my employer 6 figures.

If I write things in a high level language, I'll make the deadline. If I write in C, I won't (because it will take me at least 3 times longer).

Alternatively, my employer could hire 3 people like me to compensate (which would likely bust his budget) which may or may not help depending on the problem domain (in this case, I can confirm it wouldn't help that much).

As it turns out, the operational code that I write doesn't need to have C-level efficiency in cpu utilisation (right now, I'm writing infrastructure as code and I guarantee you we won't need to provision 1 million virtual machines per second... that would definitely bust my employers' hardware budget).

Hence, I'm not doing it in C.

nice , can i join?

Unless you genuinely care about that kind of stuff, if you're gonna put this amount of effort into something, there are easier ways to make more money, trust me.

I'm essentially a highly skilled laborer. Don't get me wrong, the kind of skill I honed (and am still honing, its a never-ending process) gives me leverage to negotiate a salary that is a lot higher than what a factory worker can aspire to, but ultimately, if you want to make crazy money (and if you put the kind of effort I put in my career into something, crazy money is not that far fetched), go into management, finance or own a company.