Please don't compare a programming language to a hammer. They're not similar in complexity, and it makes a really bad metaphor. Besides, unless you write entire applications in assembler, you DO care about programming languages.

They are similar in outcome.

Do u care about what language Amazon uses to write S3? Do u believe it is the language that make S3 succeed, or fail?

Even as a programmer, as to the other services I am calling, things I do care about involving their stability and correctness that affecting my work. I care shit about what how they write their service underneath, may it be cool or dumb.

If we're going to insist on the carpentry tools metaphor ... hammer vs nailgun; a hammer can do everything a nailgun can do and more ... but if you're banging together a timber-frame for a house the nailgun will let you get through the task far far more quickly. If you're running a business, and time is money, which tool would you want your carpenters using for this task?

If I'm running a business that lets say develops an ecommerce platform I won't be too impressed if my developers are using assembler. Java good, but if Scala gives them an edge and cuts my costs so much the better!

Yeah, so choose what works best for u, not assuming the same thing for others without understanding the specific problems they are trying to solve. Maybe they need a hammer, not a nail gun, because they work in specific environment that don't have sufficient power supply? or for the work they are doing, hammer is good enough, it is better for them to invest elsewhere.

I'm kind of making the point that in many environments you'll need both? A nailgun (battery or gas powered) for the "bulk" jobs and a hammer for the finer work. I can't imagine anyone being effective in Scala without first having a good knowledge of Java. Or at least without having Java expertise near to hand.

I used to think so too, but at work I've met programmers who code in Scala without having learned Java first, and it's enlightening. They are proficient in Scala. Whenever they touch Java code, they are upset that the niceties and modern features of Scala aren't there -- whatever we thought was reasonable in Java or "not mainstream" in Scala are exactly the opposite to them. "Why do I have to write all this needless code in Java?", they ask (for example), nicely countering the common but unwarranted notion that Java's verbosity somehow makes code more understandable to regular joes.

It's enlightening to me because a lot of what we consider "hard" or "too academic" is actually our own prejudices speaking. Of course it looks like that to us, coming from Java, but try talking to someone who learned Scala first :)

I also find it encouraging. Maybe some day we will be able to finally take advantage of the JVM without the weight of Java-the-language.

You've met programmers who know scala but not java but tell me this: Were they in an environment without java expertise "to hand"? To me scala is more akin to a skin over java. Particularly when using 3p libraries that were written in java (this ecosystem is one of scala's key advantages over other fucntionalesque programming environments). When you work with scala you'll always need some java guys nearby to deal with the "nitty gritty" details.

No, that's precisely the thing -- I've met programmers who learned Scala as their first language for the JVM (some guys I'm thinking about coded in C and PHP before that), they are pretty proficient, and they absolutely do NOT need Java programmers "at hand". It's interesting that they "get" Scala; they don't try to use it as if it was Java++ (in my experience, this attitude you describe that Scala is merely a "skin" over Java is mostly due to prior knowledge of Java). When these people eventually need to learn some Java, they do so on their own -- and curse the language!

I think we're starting to see a new phenomenon: people who don't start on the JVM with Java. It's interesting because it shows (in my opinion) that the whole "Scala is too difficult" thing was mostly a misconception. What it was truly all about was: it's hard to give up old ways of thinking :)

I honestly can't see the point in scala if you're not integrating with the the javaverse ... I'd have thought you'd be far better off cutting a clean break and using a "proper" functional language like Haskell or Go

Obviously, one big draw of Scala is that it can interoperate with Java and take advantage of existing libraries. But that's a far cry from saying it's mostly a wrapper around Java. New code, including libraries, should be written entirely in Scala. The JVM is itself a valuable target, regardless of Java-the-language.

That said, I love Haskell and would definitely enjoy using it at work, but it's a no go for "political" reasons, which is sort of reasonable even if disappointing. Scala scares management way less than Haskell!

I don't personally care what S3 is built using, but that doesn't mean the people who built S3 shouldn't care. Java, Scala, etc. are all tools with different strengths and weaknesses. Use the best tool for the job, considering the experience and capabilities of the team you have.

Just because end users don't care about the language used to implement S3 doesn't mean the developers at Amazon don't care.

_I_ don't care what they use to write their platform, because it's a black box to me. But _they_ care and spend a lot of time debating it, and that's when discussions like the one in the article are important and shouldn't be dismissed.

> unless you write entire applications in assembler, you DO care about programming languages.

That's true, but we've been in the region of diminishing returns for programming languages (I'm not talking about runtimes) for a long, long time now. To go back to the (bad?) tools metaphor, it's like saying that unless you're using nothing but your fingers, you DO care about tools. Well, yeah, but once you have a decent screwdriver, innovation in screwdriver technology -- short of a revolution -- is likely to yield diminishing returns on investment, especially as the cost of adoption remains high. Switching from, say, Java to Scala is not much cheaper (if at all) than switching from, say, assmebly language to FORTRAN or C. This does mean that the effort spent on switching languages (or screwdrivers) should decline.

I completely agree. Bricks would be a better metaphor and still completely contrived. And turns out, you can't build skyscrapers without concrete and steel :)

And just to be pedantic: When I bought my first SwissTools Insider[1], I realized that it's actually still possible to innovate in screwdriver design. That doesn't make them better for every task, but hey, that's the way it is.

[1]: http://www.pbswisstools.com/de/werkzeuge/quality-hand-tools/...

Modern skyscrapers would certainly not be reasonable to build without "strange new hammers and screwdrivers" (read: tower cranes, super-powered concrete pumps, etc).

Algorithms and data structures will be mostly the same regardless of what language you use.

Architecture will generally not be. Scala and Java encourage building applications in different ways. That doesn't mean one is better than the other (though I'd argue you could make that claim), but some problem domains are better modeled with Scala, and likely some with Java. And some programmers work better tending toward a pure, immutable, functional style, while others work better with OO, side effects, and mutable data.

While choice of language/framework may not be as important as many other choices you need to make when building an application, but it's not something that should be dismissed as irrelevant, either.

Just cause one thing is more important than another, doesn't mean the other thing is unimportant. You appear to be arguing against discussing tools, because they aren't as important as other things.

The point is that you have to care about your tools and know them really really well if you want to be able to build good software.

That's something that holds across all fields of engineering and art.