Comment by staticassertion | Hacker News Reader

Comment by staticassertion | Hacker News Reader

staticassertion

This looks like what people have been doing with Java for years, except worse because Go lacks tuning. A common approach with Java is to bootstrap your service with a maxheap based on the environment.

https://eng.uber.com/jvm-tuning-garbage-collection/

Here's another blog from Uber on JVM tuning.

Some notes:

1. They have options

2. They have a powerful gc log, no need to roll your own finalizer thing like this

3. Choosing the max heap size, which seems like what they actually wanted here, is trivial

You could say "but that's more complex", but to me it's that the JVM has far more mature features, tons of tooling and options that they could explore and adopt, and that the obvious wins are trivial to achieve through basic parameter tuning.

Further, this GOGC parameter seems to be a very weird knob. The knobs you'd run into with the JVM will often be a lot more straightforward ie: a static value for heap size vs some number that's based on a working set percentage.

I wonder if over time Go will end up with a tunable GC.

This to me looks like the classic "language 1 needs you to write more code, language 2 needs you to know and use more features", which is an endless debate.

false dilemna, if you are too mediocre to study your tech you can still write more code, it's not an exclusive property of language 1

I think the GPs point is that, starting from 0, both require effort AND are reusable.

The "debate" is usually between people who have committed to either path (creating tools or learning tools) and think that just because the marginal cost of my approach, for me is 0, it must be 0 for everyone else. Which is patently false.

I feel we can safely assume that tuning the JVM GC might cause behavior that’s pathologically slow but never actually incorrect, which is inherently safer than rewriting a bunch of production code (probably more than once) hoping for a similar result.

I don't really see it that way. The JVM has a lot of tooling already for GC tuning and a lot of simple, powerful knobs to turn. It seems not only simpler and more straightforward to tune the JVM but the ceiling for what you can do is much higher.

With Go there's one parameter, and in my opinion it's a very strange one. It also seems strange to have to (imo) hack GC metrics in using finalizers, whereas with the JVM it's simply provided to you.

Full disclosure though, I think Go is a bad language, so I'm biased.

its weirdly black and white to think a language is "bad", particularly one in heavy use

I think C is a bad language. Necessary at the time? Sure. Is everything about it bad? Of course not. But by modern standards it’s bad, and the sooner we can migrate code away from it and onto languages with fewer footguns the better.

It’s weirdly black and white to assume that just because someone thinks a language is “bad”, that opinion doesn’t have nuance.

Full disclosure, I also think go is a bad language ¯\_(ツ)_/¯

I don’t understand, you think ideally there would be 0 code written in go? You don’t see anything interesting or appealing about it? Also, calling a language bad is black and white, observing that is not black and white.

I just tried to make the point that thinking a language is “bad” doesn’t mean that there isn’t nuance behind that opinion, and so you immediately jump to the conclusion that I believe such a language has zero redeemable qualities and should be stricken from the Earth?

Hell, I enjoy writing C but I still think it’s a bad language.

If you want me to think you have a nuanced opinion, express a nuanced opinion. The opinion you expressed is unsubstantiated, and instead of substantiating it, you’re attacking me. Poor showing mate.

The grandparent was disclaiming his bias regarding his own personal dislike of golang. A full-blown point-by-point critique would have been unnecessary and inappropriate in that context.

Nothing I’ve replied here has been an “attack” on you. I simply tried—gently at first—to suggest that “x is bad” should not be equated with “x is irredeemable” which is not exactly a charitable or reasonable interpretation in the context in which that statement was originally written.

Further I directly expressed a more nuanced opinion as an example in my and you still chose to discard that nuance and interpret the opinion as black and white.

I think you’ll find that on a scale of positivity from -1.0 to +1.0, most people perceive “bad” as somewhere along the lines of “< 0.0” and not “= -1.0”.

Why do you think go is bad?

Why does it matter so much to you that I do?

I have zero interest in rehashing an argument that’s been made here hundreds if not thousands of times already, and by others far more eloquent and convincing than myself no less. Feel free to read my post history. Or simply search for virtually any golang-related post on this site. Whatever arguments you find, I probably agree with at least 80% of them.

Ok, in summary, go is bad, there’s a nuanced opinion behind it, and you won’t tell me what it is. Also, I’m bad for assuming that this is a black and white point.

Have a nice day.

I am genuinely, honestly confused as to how you cannot see how all these points are congruent with one another.

I believe go is a bad language. I have nuanced, lengthy, and detailed opinions behind that belief which stem from 24 years of software engineering, 4 years of professional experience specifically with golang, and professional experience writing, deploying, and maintaining production software using C, C++, Rust, Java, Ruby, Perl, and JavaScript. And I have zero interest in rehashing the past twelve years' worth of arguments against golang with someone who's repeatedly signaled a frustrating level of obstinance.

Whatever wild conclusions you choose to jump to from there are your own doing, not mine.

What I am getting from you are a lot of defenses and very little substance. You can claim to have nuanced opinions and no interest in discussing them, but then why are you even talking to me? What is the point of declaring that go is bad, if you are unwilling to discuss it? Is that supposed to be persuasive? I'm here to have a discussion on a discussion board. Your position is inconsistent with your actions.

Also: I don’t believe you. You have provided no evidence that you actually have a nuanced opinion, you’ve simply insisted upon it it’s possibility. And I don’t think there’s any reason I should believe you.

It feels like trying to get trumps tax returns. “They’re great returns” he insists, but he will generate all sorts of arguments to try and stop you from actually seeing them.

I don't see anything interesting or appealing about Go at all. I've read and written Go, I've watched Pike's talks on it, and I follow its development.

I could talk a lot about why I think it's a bad language, it would be hard to summarize it since I'd want to cite Pike's talks on "simplicity", articles on Go's GC implementation, discuss error handling, what I think makes a language "good", etc.

seems like they were able to tune it just fine given they wrote a library.

I didn't say otherwise, I'm just comparing the two approaches. We have a very interesting, rare situation where one company has published a blog post on GC tuning across two languages, it seems appropriate to compare them.

Reply on news.ycombinator.com