Convince your boss to let you use Scala | Hacker News Reader

Convince your boss to let you use Scala

scala-boss.heroku.com

56 pointsDanielRibeiro43 comments

Now this is just my opinion about Big Corp from experience, I'm all for Scala and have been helping people all over the world with it and Akka.

Let me first define Big Corp. A traditionally long standing business (good ole boys) type of company. A financial or insurance institution, with hundreds of employees in I.T., devs, QA, and deployment.

I've seen big corps with large teams who are entrenched in .Net or Java mention they also must worry about getting people to understand your Scala code. Hate to say it, but I've meet a lot of supposed Java programmers who didn't know what a HashMap was... in big corp...

In my experience, they just (big corp) don't give a crap. It's Java or .Net and if you don't like it... leave.

Unless that one small group going rouge and using Scala has full backing from the top you'll find yourself on a deserted island in time... which doesn't mean to say that they won't give you a shot. They will but they want 2-3x results. Otherwise why break what you have now? I'm just saying that is what I see them thinking.

I believe it has to do more with the type of project and if it has real visible impact on the bottom line to whether Scala or X,Y,Z will succeed in the big corp env. Project A does well, makes Big Corp more $$$ then they try to reproduce. Mr. VP of finance doesn't care if it's Scala or Alacs (Scala backwards) and that Scala helps you write less code. Small Corp might though as they have less resources (funds/time).

Just my opinion from interfacing with Big Corp.

I did leave (each one) after I got tired of trying to champion better ways of doing things.

Selling Scala to the PHB is an exercise in futility. What Scala needs is more success stories like the Twitter backend.

Our field is a little bit weird.

Twitter "success stories" (yes, with quote, since we never actually see the internal code of Twitter, and considering that they have many developers since that is their core) is probably not applicable to "business enterprise software". I believe there are attempts to fit solution for case X to case Y already in the history of software projects. I'm not sure if that went well.

Actually Twitter has open-sourced some of their core Scala code.

I think the problem is not just that the PHB will not see the benefits. It's difficult to hire people who know Scala -- or Haskell, Erlang, etc. Yeah, good programmers can learn it, but most companies who are hiring experienced programmers require experience in their language choice (even startups, just look at the job ads on HN).

exactly my thoughts during the presentation.

From my own experience it's actually easy to hire Haskell programmers. There's some supply, and not very much competing demand at the moment.

"Scala requires the smallest mental shift from Java". Feels to me like Groovy would be the smallest mental shift from Java, not Scala.

This is exactly what I thought.

side note: Why would anyone downvote this? What is a downvote? Does it mean it's not true? Should we downvote everything we don't agree with? Is this some kind of competition?

Cause apparently groovy's not cool?

The presentation was about how to switch to Scala and the primary consideration was "I want to do functional programming", not "I want to use something like Java, but less verbose" (which would lean more towards groovy or possibly jruby).

OT: You probably got downvoted because your comment adds nothing to the discussion—if everyone wrote "I agree" or "I disagree" (with no justification as to why) on comments they agree or disagree with the signal-noise ratio would be much lower.

The guy above me who mentioned Groovy got downvoted, not me. I only commented when I saw his relevant comment starting to grey out.

People might downvote the original comment because it adds nothing to the discussion. They might downvote it after adding the "side note" because now the post moves the discussion off topic (e.g. this post) or because meta-whining is not considered good form on HN.

Consider downvotes as an editorial tool to help you create better posts. When you see a downvote the first response should be, "what is wrong with my post?"

If it takes the thread off topic, delete your post.

If it is on topic, consider rewriting your post so that your ideas are communicated more clearly, your arguments flow logically, your position is reasonably supported by relevant details or examples, and that it does not rely on ad hominem or sweeping generalizations.

scala, clojure, JRuby and groovy each has a lot to recommend it, groovy has the performance/benchmark skeleton in the closet (not meaning you'll hit a wall, but meaning you'll have to convince your VP Eng or client or whatever that you won't hit a wall)

(and why is it not in Alioth language shootout? ATS is brilliant, but how many devs could tell you anything about it?)

http://groovy.dzone.com/articles/groovy-action-how-make-5000

> ATS is brilliant, but how many devs could tell you anything about it?

Might that be why the benchmarks game tells devs about ATS?

Sure that's probably true. As a boss I let one of my developers convince me to let him use Groovy in a small feature of one of our products. It worked pretty well for that, but everyone agreed that it wasn't worth going any further due to poor run time performance.

Only a small fraction of the code you write is performance critical. Write the app in Groovy, profile it, rewrite the bottlenecks in Java.

Or use Groovy++

Having done a lot of coding in both Groovy and Scala, I would agree that Groovy takes the smallest mental shift. However, if you require static typing, Scala is your answer.

A few things I don't like about Scala:

- Complex syntax: semicolon inference, optional parenthesis, optional braces, arbitrary infix operators, associativity rules...

- Complex semantics: classes, case classes, traits, mixins, implicits, function objects, lambdas, currying, generics, variance, view bounds, context bounds, type parameters, multiple packages, relative imports, pattern matching...

- Standard library is EXTREMELY bloated. Simple classes have dozens of supertypes and hundreds of methods. Documentation is also lacking: a lot non-obvious methods and even entire classes have no comments or scaladocs.

- Implicits are nice in theory, but in practice they lead to nasty spaghetti code which is difficult to follow. They also hide errors which would be caught be more more robust type systems. So does type inference. So does the lack of checked exceptions.

- IDE support is still poor. The problem isn't that Scala is new; it's just very difficult to build a great IDE for such a complex language.

- Ugliness resulting from JVM compatibility: null, ClassManifest, Function/Tuple classes, nasty mangled names...

- Compilation is very slow, with or without fsc. Compiler is more stable than it was a few years ago, but it still does funny things and occasionally crashes.

- Huge memory overheads and lousy performance resulting resulting from all the indirection.

Yep, I've been using it for a couple months and encountered all these issues.

WTF is the point of it all?

Or you could switch bosses and come work at Yammer where we use Scala to ship infrastructure services every week: http://jobvite.com/m?3m785fwq

How tragically ironic that this presentation doesn't even work on my BigCorp laptop restricted to IE8.

"Scala would have a positive impact on recruiting"

I'll bet this is a big Yes for Scala, Groovy or Clojure compared to Java.

Where are good places to look for people who want to work in Scala? My startup is currently all Ruby, but we're thinking Scala is a good candidate for some performance-critical pieces.

Scala mailing lists, IRC, functional programmer meetups.

Try hanging out on IRC (e.g. #scala on freenode). Either ask there directly, or ask for pointers.

It's looks like VMWare might be a good place.

A few others: http://getvouched.com/jobs?term=&terms=%2C6d81ef86-fe56-...

The argument is not convincing because it fails to demonstrate that Scala will make the boss's quarterly bonus larger.

If you're boss can't imagine how "delivering better software cheaper and more quickly" could get him a larger quarterly bonus, you need a better boss.

"What do they (bosses) want?"

Sometimes they want software that is a good selling point. A language can be 100 times better but if the customer never heard about it it can be hard to sell. Some customers just want there product written in .Net or Java.

> Some customers just want there product written in .Net or Java.

Those people are 'clients' rather than 'customers'. The former are people who mostly likely are paying you to create something specific. The latter are purchasing a program, or access to a program on the web, and unless it's something that integrates with other bits of IT infrastructure, likely don't care a bit what it's written in as long as it works.

One of the things I'm loving about working on LiberWriter is that people don't care in the slightest what it uses as long as it does what they need. Night and day from regular battles, discussions and intrigues in the consulting world over tech choices.

Luckily you can say your Scala program is written "on Java" if not "in Java" in your marketing literature.

That's how I've been getting Groovy in the door

Good for you!

Good bosses also want happy employees.

True, but you need customers as well.

"Always use the best tool for the job".

Also, switching from Java to Scala without leveraging it's main features (using Actors, functional programming) seems like switching for no good reason, other than the fact that Scala is trendy nowadays.

The article is suggesting that Scala is a better tool than Java for jobs similar to his. He spends about 80 slides arguing why (e.g. less code, business logic doesn't get lost in implementation detail, more maintainable).

Those are all benefits of Scala over Java, so what makes them less important than Actors and functional programming? (Notwithstanding the fact that some of those benefits are because of gradually introducing the more accessible functional programming constructs - like filter and Option.)

That's like saying you shouldn't use Ruby unless you plan to write lots of DSLs and monkeypatch everything. Some people do those things with Ruby, and it's good at them, but it does other things too.

It's a bit disingenuous to talk about Scala without pointing out that the IDE tools are, from my understanding at least, fairly poor when compared with Java's tools. That doesn't mean you shouldn't use it, but if you're trying to convince your boss to use something, totally glossing over a fairly large negative is dishonest. Do the productivity gains from less verbose, clearer code? Probably, but the point is it's not just Java++; it's Java + some very important things - some very important tools, which is a totally different equation.

Can't navigate the presentation because I disabled the arrow keys with a user script the day Google hijacked them. I hope this trend of interfering with standard browser functions ends soon.

While off-topic, this is relevant. I don't like it either, and I think there are many others who would agree. Have an upvote.

Touch devices seem to be simillarly impaired. Seems like someone was being a little to clever.

Reply on news.ycombinator.com