News for hackers. Stuff that matters.
Re:Rust required to build Gecko by steveklabnik on Friday December 16, 2016 @11:20AM (#13194265)
> I find it fascinating that lessons aren't being learned from build systems which are built in the enterprise

What specifically makes you think lessons aren't being learned?

> honestly, I'm disappointed one of those two wasn't just adopted by the Rust community - unclear but I think they didn't know about them?

Bazel had its first release in March 2015, and wasn't even considered beta ready until later that year. Pants had its 0.0.17 release, the very first public one, in July of 2014.

Cargo was announced in March 2014, earlier than both.

> Why only those two?

Well, those two are necessary. Other tooling is nice, but not the bare minimum needed. We need to stay focused, but more on that later.

> Why not learn from the whole Java build pipeline?

I have not dealt with the Java-specific tools you're talking about in a very long time, but pipelines have advantages too. A few slow tools doesn't mean the whole concept is bad.

> Why is clippy, like Findbugs, separate from the compiler?

Rust cares a _lot_ about stability. Adding new lints is extremely difficult, because they can break people's builds. Clippy being a separate tool means that they can iterate on lints at a different release cadence than the compiler itself, and aren't bound by our mega strict breaking changes policies.

> In the enterprise, you work with great developers and you work with not so great developers.

I assure you, it is very much the same in open source. ;)

So, we do very much care about Rust's usage in enterprise scenarios. For example, Mozilla in many ways feels exactly like an enterprise customer of ours. It was a huge amount of technical (and social) work to get to this point, and a lot of that involved negotiating how the builds would actually work. Our perspective is that Cargo is a world-class tool for building code, but that's only possible because it's focused on _Rust code_. So the strategy should be, how can we best integrate Cargo with these more generic build systems? We've already taken a number of steps forward on this front, and there's more to come.

Most of these features lacking in Cargo are due to a lack of some kind of champion to lay out the specific needs; we don't want to just build features and hope they're useful. In the past, we've had several people come to us with "here's my issue getting Rust in my workplace", and that experience has been valuable. We're a small crew, we need to focus in order to keep shipping quality stuff. We don't have the ability to just toss some people on it, for whatever value of it, and get it done.

Re:Rust required to build Gecko by yazaddaruvala on Friday December 16, 2016 @04:48PM (#13197186)
https://news.ycombinator.com/item?id=13197159

> Cargo was announced in March 2014, earlier than both.

Thats fair, and a part of what I meant about "they didn't know about them", because working for Amazon right out of college, I've been using things like Bazel/Docker/Travis for 5+ years now.

I actually meant it as a discredit to people who work in the industry for not exposing these tools earlier. I did a poor job of communicating that (it was after all like 3 am :P).

Re:Rust required to build Gecko by steveklabnik on Friday December 16, 2016 @05:49PM (#13197617)
It's all good! Just setting the record straight :)