[zellyn]

> Can you give an example? I guess I fail to understand why a statically typed language would choose to forgo all of the advantages generics provide. Does it have something to do with Goroutines?

Like I said, the objections are practical, not philosophical:

They published four past design docs for generics in Go that simply didn't pass technical muster (https://github.com/golang/proposal/blob/master/design/15292-...)

Also, rsc stated he plans to understand generics better in 2017: https://research.swtch.com/go2017#generics

Also, bradfitz said recently on the GoTimeFM podcast that doing Generics and Go2 together makes sense.

> I didn't necessarily mean their marketing budget. I should probably edit that. I meant that anything they do is news.

I feel like the comments from others about Dart give the lie to this one. Go is fantastically, dramatically, massively more popular than Dart, which is also from Google.

[unscaled]

> Like I said, the objections are practical, not philosophical

Russ Cox and Rob Pike keep repeating this mantra, but I don't buy it. No, I don't believe they object the idea of generics itself. And yes, generics pose practical complexities and a plethora of issues that have to be resolved.

But so does every other language feature. The features you choose to add reflect your philosophical values and priorities.

Go chose to bake some hitherto very niche features that could have been as libraries into the language. Channels in particular, are a language construct only because Go doesn't allow operator overloading and generics for user types. But channels get them, because channels are demonstratively important for Rob Pike[1].

There's nothing wrong with that of course, but that's a philosophical decision. Why channels can avoid the vagaries of interface{} boxing, but not sets, linked lists or queues?

The reality is that every modern statically typed language except Go has generics, and they all implemented them very well. The ML languages and Ada were already doing it in the 80s, and OCaml and Eiffel managed to combine generics and polymorphism back in the 90s.

It seems to me that when originally designing Go, Pike, Griesemer and Thompson just didn't think generics are worthwhile enough for the effort it takes to properly research them.

When Go was started, its authors mostly looked to languages which implemented generics later in their lifecycle (namely Java and C++) and their implementations suffered from problems due to other, rather obvious, design defects: https://research.swtch.com/generic

I'm happy to see that this attitude is changing, and other languages are looked at.

[1] https://swtch.com/~rsc/thread/

[zellyn]

> The reality is that every modern statically typed language except Go has generics, and they all implemented them very well. The ML languages and Ada were already doing it in the 80s, and OCaml and Eiffel managed to combine generics and polymorphism back in the 90s.

> It seems to me that when originally designing Go, Pike, Griesemer and Thompson just didn't think generics are worthwhile enough for the effort it takes to properly research them.

Are you arguing that it's easy, and they just haven't done it? Or are you arguing that it wasn't important enough at first, and so the implementation developed in directions that preclude straightforward implementations now? I can believe the latter, but after reading the multiple, detailed proposals from Ian Lance Taylor, I don't believe the former.