Hacker News Reader

Comment by shawn | Hacker News Reader

Go is still one of the most difficult and counter intuitive build systems ever devised. Usually you’d expect a project to build out of the box. Go seems to take a stance of “Haha nope, get ready to figure out which version of the standard library this 4 month old project was written against, then have fun trying to update it to the latest.”

I like the language though. Their green threads are still the best.

View on HN

From Go's 1.0 compatibility guarantee[0]:

> It is intended that programs written to the Go 1 specification will continue to compile and run correctly, unchanged, over the lifetime of that specification. [...] Go programs that work today should continue to work even as future "point" releases of Go 1 arise (Go 1.1, Go 1.2, etc.).

I've found this to be upheld in practice, especially so for the standard library. I have yet to come across a case where I was concerned with which "version" of the standard library some project was written against. Are you perhaps talking about non-stdlib packages?

[0]: https://golang.org/doc/go1compat

My experience has been the same, so it’s hard to understand where shawn is coming from. There have been impressively few breaking changes to the Go standard library since 1.0.

Even when bringing in a number of third-party packages, the only trouble I’ve had building Go projects has been in Windows (which is often treated by Gophers as a second-class platform).

> Go is still one of the most difficult and counter intuitive build systems ever devised.

I'm currently working on building all the coreclr/corefx repos for .NET Core. Their build system is incredibly convoluted.

For what it is worth, I agree 100% and dependency management tools like Glide have been indispensable for me.

Can you tell at least one specific example where they broke the standard library?

It figures that my critique would be undermined by the choice of a single word.

Look, maybe the stdlib is rock solid. But every time I try to build a Go project, `go build` finds ten ways not to work.

Case in point: https://github.com/cretz/tor-dht-poc

Try building this. I have no idea what's wrong with it, but it doesn't work. It's cool tech, I want it, but I don't want to immerse myself in the world of hurt^WGo just to get it.

I did however at least attempt to fix the problem: https://github.com/cretz/tor-dht-poc/pull/2

The 'problem' is that the authors did not use dep (soon to be modules) and add the Gopkg.[toml|lock] files.

Or that such tools aren't shipped and part of the standard Go experience.

Fortunately that's changing.

ant, maven and gradle are not shipped with the JVM, yet the java community figures out how to do a build system somehow.

in fairness, Java has an excuse that it’s ancient in language terms and expectations were different then.

Go, not so much - it came out in 2012. Pip and CPAN and Gem had been around for years. Go’s lack of a sane packaging solution this far into its lifetime remains a serious weakness of an otherwise solid platform.

One could argue that golang is a solid platform because they focused on the platform first, instead of the package management. It is all about priorities.

Java doesn't bundle and market a `java get` command that works fine until it suddenly dosn't.

The `go get` command was never marketed as a versioned package management tool. `go get` also works just fine if you point it at forked repositories that you control.

> The `go get` command was never marketed as a versioned package management tool.

Because the Go team kept pushing the ridiculous idea that versioning doesn't matter.

> `go get` also works just fine if you point it at forked repositories that you control.

And at that point, what value does `go get` add?

It has nothing to do with Go the language itself. There are tools for that and lazy authors have the same problem with other languages also.

Reply on news.ycombinator.com