Hacker News Reader

Comment by kryptiskt | Hacker News Reader

To me it looks like simple, clear examples of potential issues. It's unfortunate to frame that as "crapping on Go", how are new Go programmers going to learn about the pitfalls if all discussion of them are seen as hostility?

Like, rightly or wrongly, Go chose pervasive mutability and shared memory, it inevitably comes with drawbacks. Pretending they don't exist doesn't make them go away.

View on HN

Go famously summed up their preferred approach to shared state:

> Don't communicate by sharing memory; share memory by communicating.

Which they then failed to follow, especially since goroutines share memory with each other.

Threads share the same memory by definition, though. When you isolate these threads from a memory PoV, they become processes.

Moreover, threads are arguably useless without shared memory anyway. A thread is invoked to work on the same data structure with multiple "affectors". Coordination of these affectors is up to you. Atomics, locks, queues... The tools are many.

In fact, processes are just threads which are isolated from each other, and this isolation is enforced by the processor.

Goroutines aren't posix threads. They could've lacked shared memory by default, which could be enforced by a combination of the compiler and runtime like with Erlang.

Who is "they"? This isn't Rust. It's still up to the developer to follow the advice.

Anyway, I would stop short of saying "Go chose shared memory". They've always been clear that that's plan B.

Go's creators said "Don't communicate by sharing memory", but then designed goroutines to do exactly that. It's quite hard to not share memory by accident, actually.

It's not like it's a disaster, but it's certainly inconsistent.

I don't think allowing developers to use their discretion to share state is "certainly inconsistent". Not sure what your threshold is for "quite hard" but it seems pretty low to me.

Goroutines could've lacked shared memory by default, requiring you to explicitly pass in pointers to shared things. That would've significantly encouraged sharing memory by communicating.

The opposite default encourages the opposite behaviour.

Go is a bit more of a low level language compared to actor languages where the language enforces that programming model. I think the point of the slogan is that you want to make the shared memory access an implementation detail of your larger system.

Concurrent programming is hard and has many pitfalls; people are warned about this from the very, very start. If you then go about it without studying proper usage/common pitfalls and do not use (very) defensive coding practices (violated by all examples) then the main issue is just naivity. No programming language can really defend against that.

You are completely dismissing language design.

Also, these are minimal reproducers, the exact same mistakes can trivially happen in larger codebases across multiple files, where you wouldn't notice them immediately.

The whole point of not using C is that such pitfalls shouldn't compile in other languages.

> Pretending they don't exist doesn't make them go away.

It's generally assumed that people who defend their favorite programming language are oblivious to the problems the language has or choose to ignore these problems to cope with the language.

There's another possibility: Knowing the footguns and how to avoid them well. This is generally prevalent in (Go/C/C++) vs. Rust discussions. I for one know the footguns, I know how bad it can be, and I know how to avoid them.

Liking a programming language as is, operating within its safe-envelope and pushing this envelope with intent and care is not a bad thing. It's akin to saying that using a katana is bad because you can cut yourself.

We know, we accept, we like the operating envelope of the languages we use. These are tools, and no tool is perfect. Using a tool knowing its modus operandi is not "pretending the problems don't exist".

While I agree with you in principle, there is a small but important caveat about large codebases with hundreds of contributors or more. It only takes 1 bad apple to ruin the bunch.

I'll always love a greenfield C project, though!

> Using a tool knowing its modus operandi is not "pretending the problems don't exist".

I said that in response to the hostility ("crap on Go") towards the article. If such articles aren't written, how will newbies learn about the pitfalls in the first place?

Reply on news.ycombinator.com