Scala - is it going to consign Java to legacy code

richie murray

I hope that this is posted in the right place. I have been researching Scala as several of my clients are making the move over to it. I work in recruitment (boo I know) and when talking to clients and candidates I like to know what I am talking about. We don't all just bluff our way through.:)

I do not code myself so have no hands on experience with Scala but it does seem to be very powerful and for want of a better description seems to be much cleaner and require less code than Java. Basically more powerful and requires less man hours to get everything up and running.

I tried to find some info on here and may have been looking in the wrong place so please move\merge this thread if needed. What I wanted to ask is if any of you guys who are using it in a professional environment are finding it to be what it is billed to be or if it will be another language that didn't quite make it. I have read quite a bit on its potential but as with everything in life things can fall short of their billing.

In short, how good is it in practice and will it actually replace Java?

srsly78

It's just a functional language that happens to compile to the same bytecode. Still uses the JVM.

It will replace java just like lisp replaced everything else (it didn't).

ChRoMe

Will it replace Java? not a hope.

Marko_Polo

I think it's possible, but we're still getting out of the stone age of VB6 applications and just crawling into the OOP age with a lot of systems which take years to develop, and they don't even teach Scala in Irish colleges (at least I highly doubt it).

I think it would likely take several years for academia to accept it, another 4-5 years for enough graduates who would know it, and then years more to adopt it fully into the mainstream. I honestly think even if it does replace Java it will take a long time, at least 10-15 years.

Just remember 15 years ago people were talking about C++ replacing C etc. Where are we now, C and C++ co-exist and we see fluctuations between which is more in demand.

On a personal note as a Java dev, I do plan on digging into Scala over my xmas break....just to see what all the fuss is about

srsly78

Marko_Polo wrote: »

I think it's possible, but we're still getting out of the stone age of VB6 applications and just crawling into the OOP age with a lot of systems which take years to develop, and they don't even teach Scala in Irish colleges (at least I highly doubt it).

I think it would likely take several years for academia to accept it, another 4-5 years for enough graduates who would know it, and then years more to adopt it fully into the mainstream. I honestly think even if it does replace Java it will take a long time, at least 10-15 years.

Just remember 15 years ago people were talking about C++ replacing C etc. Where are we now, C and C++ co-exist and we see fluctuations between which is more in demand.

On a personal note as a Java dev, I do plan on digging into Scala over my xmas break....just to see what all the fuss is about

Terrible example seeing as C++ did mostly replace C. C only gets used in embedded systems these days.

Also wrong about acceptance in academia, all these weird languages come out of academia. Academics use all kinds of ****ed up languages - industrial acceptance does indeed take a loooong time. I can look back at all the esoteric stuff I got taught in Comp Sci - Eiffel for example. This was a big academic language that never took off in industry.

dazberry

Marko_Polo wrote: »

and they don't even teach Scala in Irish colleges (at least I highly doubt it).

I took a programming paradigms module in GCD in 2010 and it was all through Scala, still is apparently. I haven't looked at it since, but dug quite deeply into it at the time and was very impressed.

D.

stevenmu

Scala does look impressive. And there would certainly be a good argument that it is better than Java. At best though I think it could only be said to be an evolutionary improvement, not a revolutionary one. And I think the niches that it fills are programmer driven ones, not business ones, i.e. programmers will like the extra features it has, but it will be hard to make a business case based upon them.

If you look at the languages that have succeeded over the last decade or so, they have all offered something compelling to make people use them;
Java: Offered the promise of true platform independence.
Javascript: For better or worse it was the only non-extension client side game in town long enough that people finally gave in and made the best of it (and when they did it turns out to be quite good).
.Net: Became the development method of choice for Microsoft stuff, offered a feature rich web framework at the right time, and was orders of magnitude better than VB.
PHP: Offered quick, simple and efficient web development, and has been improved with frameworks.

On the other hand, I don't think Scala offers anything really compelling that you can use to justify it to a manager. Especially one with an existing Java code base. And I'd imagine future Java versions will incorporate some of Scala's advantages.

On top of all that I think Python, Clojure etc have managed to corner a lot of the "like Java but better" market.

I can imagine it being used in some start-ups, and making some inroads into more cutting edge companies where individual teams have some flexibility. But without something truly compelling to drive it forward I can't see it becoming a de-facto replacement for Java any time soon.





(Btw it's great to see someone in recruitment taking an actual interest in the industry. So many don't, and they give the rest a bad name.)

richie murray

Thank you very much for your views guys and I really appreciate it. I cannot thank the individual posts as I dont seem to have the option. I guess I need a certain number of posts before that unlocks.

I will keep researching and as I get my briefs from the heads of development within my clients to get my briefs I am sure that I will get some more insight. I will post up why they have chosen to use it incase it is of interest to anyone here.

If anyone else comes across anything else or has any views I would be greatly interested in hearing them.

Feathers

richie murray wrote: »

I do not code myself so have no hands on experience with Scala but it does seem to be very powerful and for want of a better description seems to be much cleaner and require less code than Java. Basically more powerful and requires less man hours to get everything up and running.

It's not a case of, you write less lines of code, so therfore you need less man-hours — Scala has a more concise syntax, which can make it easier to read, but written badly it could be even harder to debug with custom DSLs overriding everything & especially with a younger set of tools to Java.

If you work in Tech recruitment, you'll know how well-paid the positions for multi-threaded app devs are; functional languages come at this problem in a different way, & I think it's much easier for your average developer to work with. The need to make use of parallel programming is going to be on the rise, (as that's how computers are getting faster these days as opposed to clock speeds), & you can see the use of Scala in financial places & firms dealing with 'big data' like Twitter. But Java 7/8 will close out this gap somewhat.

The JCP seems to view other languages on the JVM as a chance to see what works & doesn't — merging back in the best to Java, so Scala being a 'Java killer' is unlikely. I think it will find favour in companies that are happy to be ahead of that curve though.

Marko_Polo wrote: »

I think it would likely take several years for academia to accept it, another 4-5 years for enough graduates who would know it, and then years more to adopt it fully into the mainstream. I honestly think even if it does replace Java it will take a long time, at least 10-15 years

Just to point out that Scala is written by Martin Odersky— a university professor who wrote the current Java compiler too, so would like to assume he knows the JVM well enough. It's put in it's time playing pubs & clubs, it hasn't just gotten the Louis Walsh ticket to stardom

stevenmu wrote: »

On the other hand, I don't think Scala offers anything really compelling that you can use to justify it to a manager. Especially one with an existing Java code base. And I'd imagine future Java versions will incorporate some of Scala's advantages.

On top of all that I think Python, Clojure etc have managed to corner a lot of the "like Java but better" market.

I can imagine it being used in some start-ups, and making some inroads into more cutting edge companies where individual teams have some flexibility. But without something truly compelling to drive it forward I can't see it becoming a de-facto replacement for Java any time soon.

As a developer, I like to sketch in higher level languages that don't get in the way of the thought process as much. Scala is strongly typed, but hides this away from you — the compiler will only complain if it can't infer your meaning unabiguously from your code.

From a business point of view, one major plus is that the code still runs on the JVM & has full interoperabilty will Java. If you wanted to move to Python, you've to make sure you're architecture supports what you're running, the performace of the code is as good & endpoints for change over between languages are well-defined.

With Scala, you can copy & paste the code from Foo.java to Foo.scala & your IDE will do a first pass at coverting between the languages. You're not by any means writing good Scala, but it will compile & run and perform as well as equivalent Java code — this is a great boon to a business looking to dip its toe in the water of a new language.

At the next stage up from that, I think it's main selling point is that it's both functional and OO — you can structure your data in an OO way, when that makes sense, but have your utilities performing functional operations in an immutable way. You can of course write code to do this in Java, & this will be easier after Java8, but from the little I've used Scala, I find that the conventions in the language channel you into thinking in an immutable way much more, which is important for easier distributed computing.

stevenmu wrote: »

(Btw it's great to see someone in recruitment taking an actual interest in the industry. So many don't, and they give the rest a bad name.)

Definitely +1

richie murray

Feathers thank you for putting a detailed answer together. You are indeed correct the guys who I am working with are all involved in "Big Data".

Will be interesting to see how this all develops and I seem to have picked a few things up since parroting off some of what I picked up here gave a softer opening to the hiring managers who were happier to spoon feed me through the reasons they are going down the Scala route.

Lots more info to give to my candidates now so should be giving them facts rather than selling snake oil.

In short thank you to all who commented. Greatly appreciated.

Sparks

srsly78 wrote: »

Terrible example seeing as C++ did mostly replace C. C only gets used in embedded systems these days.

/blink
Systems programming language only used for systems programming?
Also, it's not just embedded systems, it's pretty much all the low-level stuff; so the linux kernel, every other unix kernel, most unix tools, the majority of device drivers, and still quite a few applications, not to mention python, php and perl (all of whose compilers are written in C).

C hasn't gone away - C's just been a bit busy running your world.

Also wrong about acceptance in academia, all these weird languages come out of academia. Academics use all kinds of ****ed up languages - industrial acceptance does indeed take a loooong time. I can look back at all the esoteric stuff I got taught in Comp Sci - Eiffel for example. This was a big academic language that never took off in industry.

Okay, that's two dumb assertions in one post. That's pretty much the point where you should stop. Eiffel and Smalltalk and Pascal and dozens of other "academic" languages might not have gone on to be used widely in industry, but they pioneered approaches and features that went on to become the core of other languages like Java and C++ that did become widely used in industry. That's the entire point of academia - to do the unprofitable blue-sky research that industry can't afford to do and to thereby build the next tools we'll use in industry.

And last time I looked, by the way, Professor Bjarne Stroustrop wasn't well-known as an industrialist and Ada wasn't widely considered to be an example of excellence in programming languages

Sparks

richie murray wrote: »

In short, how good is it in practice and will it actually replace Java?

Haven't used it and Feathers covered it better anyway; but the "will language X replace language Y" question is an old one and the answer is always "no". There's still absolutely enormous codebases out there in Cobol and FORTRAN that are not merely actively maintained, but are actively running critical systems for society (ie. banks, logistics networks, that kind of thing, where you can't just turn the system off without disrupting life for everyone). And they're still a business concern - large databases like Oracle and DB2 still provide Cobol APIs in their latest versions this year, for example.

Just because it's legacy code doesn't mean that it's not worked on - and there's good money in it.

And this year's TIOBE list is worth reading: in the top place is C, and in the top 20 you have assembly, Ada, Lisp and two versions of Pascal - all languages from last century And that's without bringing up Objective-C which is now all popular and shiny and new-looking as the iPhone/Mac programming language of choice, but which is actually much much older and if I recall correctly, predates C++ by a few months.

srsly78

The Tiobe list is load of ****e and the only reason it gets so much press is there is no other benchmark. All it does is show what languages people are searching for tips about on google. So it shows FOTM (flavour of the month) but not industry adoption.

And you call my assertions dumb, but then repeat them? Yes the academic work was pioneering, but we still don't use "pioneer" languages. As a professional developer I want to earn money, not waste my time ****ing around with some academic nonsense - but here is where our backgrounds diverge I suppose. Feel free to continue missing the point of my posts.

I should add that in the 90s my lecturer was a huge Eiffel fanboi and told us all that it was the next big thing etc etc. Also I'm stuck maintaining a C/C++ legacy codebase in my current contract, so you are god damned right I'm mad

conor.hogan.2

Tiobe and langpop place C as most used, as does any one with any basic reasoning.

Sparks

srsly78 wrote: »

The Tiobe list is load of ****e

Ah, reasoned argument.

and the only reason it gets so much press is there is no other benchmark

...except that it's not.

• PYPY
• LangPop (for as long as it stays up)
• A rather neat dive into one job site's stats

There are more, but they're all basicly doing the same thing as Tiobe; they're trying to judge the popularity of a language by diving into what is not fantastic data to begin with, but which is the only data that will ever be available to work on since most companies aren't going to dump internal information into the public domain en masse. So they'll never get precise ground truth; but what they can do is give indications and with certain marketing-written blurbs aside, that's all any of them are claiming.

BTW, it's moderately amusing that some of the best criticism (as in, the most logical and consistent criticism) of Tiobe says that Tiobe is wrong because there's not enough academic research into some of the languages on its list. I thought you'd find that funny given your love of academia.

Also, from the point of view of a recruiter, TIOBE (and its ilk) is about ten miles better than nothing at all. And from the point of view of a student, you don't have a choice in what language the course is run in; and you should be doing any extracurricular work in whatever you find interesting or don't yet know (because that's the point of being a student - you can't go off and spend client time mucking about learning a new language, so you do it in college or on your off-time, and as you will learn after studenthood, off-time is something you should bill at the same rate as on-time, more when you start a family, and so your mucking-about time still gets severely cut after you graduate).

Yes the academic work was pioneering, but we still don't use "pioneer" languages.

Really?
C was a pioneer language, don't forget. So was C++. So was Java, at least in the beginning. Some pretty major applications out there are written in other pioneer languages, and that's not counting those written in domain-specific languages. Hell, go walk round a hospital's IT department; you stand good odds of running into languages you'd imagine you'd only ever encounter in old books. You're falling into the mindset that believes that if the blogosphere isn't talking about it, it isn't happening. Truth is, there's more code running in the world on 4- and 8-bit machines right now that was written in assembler than there is of anything else. Just because we don't think about it much doesn't mean it's insignificant, or unworthy of study, or that you can't earn a decent living doing it.

As a professional developer I want to earn money, not waste my time ****ing around with some academic nonsense - but here is where our backgrounds diverge I suppose.

I'm sorry, your professionalism has blinded me with awe, especially that bit where you described major commercial success stories like C++, Delphi, C, and a dozen or more others; not to mention the experiments that made dozens more commercial successes possible in the first place; as academic nonsense.

I should add that in the 90s my lecturer was a huge Eiffel fanboi and told us all that it was the next big thing etc etc.

In the 90s?
You mean right before half the design of Eiffel got rolled into Java and released as the "Next Big Industrial Language"?

Also I'm stuck maintaining a C/C++ legacy codebase in my current contract, so you are god damned right I'm mad

Ah, professionalism, where "legacy" (and presumably "paid work during an economic depression" means "not cool enough for me to be stooping to"

Colonel Panic

Since when did legacy become a bad thing? I'm banging out some amazing C++ these days on a 15 year old code base.

ChRoMe

Colonel Panic wrote: »

Since when did legacy become a bad thing? I'm banging out some amazing C++ these days on a 15 year old code base.

It implies that any future work on it is going to be maintaince rather than greenfield, which makes it seem boring to some people.

Sparks

ChRoMe wrote: »

It implies that any future work on it is going to be maintaince rather than greenfield, which makes it seem boring to some people.

Which is why whenever people say "oh, you have to be passionate about your work to be great at your job", I wind up straining eyemuscles from all the rolling Here we are, society dependant on legacy software systems to an insane degree (think about it - hospitals, food logistics, water and power management systems, banks and the rest of the financial sector, every bit of embedded software out there in everything from your anti-lock brakes to your oven to your ambulance's AED, there's an enormous amount of code out there running society-critical systems) and if that code's not maintained properly, Bad Things (and sometimes Really Bad Things) can happen. And that's not just good honest work, it's good honest important (and usually challanging) work; it's just not clean and simple work, so nobody talks it up in blogs or tweets, and so nobody gets "passionate" about it.

Meh. Give me professional over passionate any day of the week and twice on mondays please.

ChRoMe

Sparks wrote: »

Which is why whenever people say "oh, you have to be passionate about your work to be great at your job", I wind up straining eyemuscles from all the rolling Here we are, society dependant on legacy software systems to an insane degree (think about it - hospitals, food logistics, water and power management systems, banks and the rest of the financial sector, every bit of embedded software out there in everything from your anti-lock brakes to your oven to your ambulance's AED, there's an enormous amount of code out there running society-critical systems) and if that code's not maintained properly, Bad Things (and sometimes Really Bad Things) can happen. And that's not just good honest work, it's good honest important (and usually challanging) work; it's just not clean and simple work, so nobody talks it up in blogs or tweets, and so nobody gets "passionate" about it.

Meh. Give me professional over passionate any day of the week and twice on mondays please.

LOL, did I touch a nerve

I agree there is a lot of important work in those legacy systems, I think there is a danger in that career choice of having your skills stagnating though.

Sparks

ChRoMe wrote: »

LOL, did I touch a nerve

Little bit
Getting a bit sick of seeing ****e like this:



being passed off as being the norm in our industry by a tabloid blogosphere and having bad work practices become expected.
When you start seeing job ads actually say "we want a rock star developer" in those words, you know the joke's gone too far

I agree there is a lot of important work in those legacy systems, I think there is a danger in that career choice of having your skills stagnating though.

Well, true...
...'course since we're not going to replace a few billion lines of mission-critical Cobol code anytime soon, you could also call that job security

Hell, the USAF just dropped their latest attempt to modernise their logistics systems after spending a billion dollars on it because they realised they were going to have to spend another billion just to get a system that did a quarter of what the old one did - replacing these legacy systems isn't a simple task, they're old and crufty because they've been around working on difficult problems with lots of handled edge cases for decades...

Colonel Panic

Legacy systems, like modern stuff with fashionable frameworks can be written well, or not. I think it's just a common euphemism people use when the are working on old stuff and afraid of being looked down on by their peers. It's a shocking attitude in software development, referring to X technology as Y expletive with zero explanation. Who cares? You're programmers, get the job done and don't moan about it.

I dunno how banging out code with the latest Web UI, ORM and DI framework is any better or worse than working on some old MFC application. I see the same mistakes and bad design choices in both by programmers who think they're rockstars.

There's plenty of time for passion and being professional too. I don't give a crap about the applications I make in work, but I really enjoy doing them right and a lot of what I learn from my personal projects filters through to things I do professionally.

Tim Robbins

Good article of future of Java here.

Anyone know what companies are doing Scala in Dublin?

From what I have heard: Gilt, Workday and Lab49.

I think Java is dead. You only have to look at how long it takes to develop anything in Java, how much code you have and how messy it ends up getting. Scala is better back end language. And a lot of Java specs are bordering on irrelevant (EJB, JSF) for example. The thing is to be a really good Scala programming I think you'd need to know Java very well first and know it's limitations and then how Scala makes things better.

I would not like to see Scala code from someone who was not an expert in Java (or some other OO language).

httpete

Tim Robbins wrote: »

And a lot of Java specs are bordering on irrelevant (EJB, JSF) for example.

Can you expand on that point a bit?

slyph

Tim Robbins wrote: »

I would not like to see Scala code from someone who was not an expert in Java (or some other OO language).

I hope what I come up with for my Final Year Project isn't so bad....

It will be a good learning experience for me anyways, some very interesting stuff.

I'd be interested in a list of irish companies using Scala also, for potential graduate jobs.

conor.hogan.2

slyph wrote: »

I'd be interested in a list of irish companies using Scala also, for potential graduate jobs.

If this is a project that will be expo'd then they will likely find you, if many of them exist.

Tim Robbins

httpete wrote: »

Can you expand on that point a bit?

Even if you wanted to use Java on your back end you'd be mad to use EJBs. Use Spring instead and write your own framework.

EJB is problematic because they you are tied to your App Server. It's a 2002 approach. JSF is out because the fat client using JavaScript is the way to go now.

There are very few companies in Dublin with a good technology stack. Most IT work is working with legacy stuff.

Tim Robbins

slyph wrote: »

I hope what I come up with for my Final Year Project isn't so bad....

It will be a good learning experience for me anyways, some very interesting stuff.

I'd be interested in a list of irish companies using Scala also, for potential graduate jobs.

I wouldn't expect a graduate to know that much.
You really need 5 years working on real projects and then it starts to plateau out. A bloke with 6 years is as good as a bloke with 12 / 18 years experience. You learn a lot the first 5 years and then you kinda of know what level the person is at.

College is about theory. The real world where you get tonnes of bugs and they have to be turned around quickly, rapidly changing priorities and working in teams is a different game.

Sparks

Tim Robbins wrote: »

I think Java is dead.

I really don't think any computer language dies off that quickly.
For example, we still have a huge installed and depended-upon codebase out there in Fortran and COBOL; we have hospitals run on MUMPS and design shops dependant on Lisp (to say nothing of the many, many emacs people who're now saying something rude about how Lisp is very healthy, thanks ). Hell, shell scripts run half the computing world behind the scenes and I don't think you could claim shell was exactly cutting-edge.

Simple fact is, the language, commercially speaking, is immaterial. And so long as the system remains in use and satisfies the commercial requirements, then that language is going to be alive.

There's not an engineer out there for whom building new things isn't as addictive as cocaine but that doesn't mean that legacy work is in some way inferior or a refuge for the incompetent or dull. Hell, technically every NASA probe is a legacy system given the hardware and software platforms they run on

Tim Robbins

Sparks wrote: »

I really don't think any computer language dies off that quickly.
For example, we still have a huge installed and depended-upon codebase out there in Fortran and COBOL; we have hospitals run on MUMPS and design shops dependant on Lisp (to say nothing of the many, many emacs people who're now saying something rude about how Lisp is very healthy, thanks ). Hell, shell scripts run half the computing world behind the scenes and I don't think you could claim shell was exactly cutting-edge.

Simple fact is, the language, commercially speaking, is immaterial. And so long as the system remains in use and satisfies the commercial requirements, then that language is going to be alive.

There's not an engineer out there for whom building new things isn't as addictive as cocaine but that doesn't mean that legacy work is in some way inferior or a refuge for the incompetent or dull. Hell, technically every NASA probe is a legacy system given the hardware and software platforms they run on

Well yeah. It depends on how innovative the company are. Of course.
Java will be around for a long time. But if you were starting a new proj, or performance / scalability is really important you wouldn't use it. This means it will eventually be consigned to a certain box.

It is easy to give examples of technologies coming in and staying around for a long while. But it is also possible to give examples of technologies that were fashionable and than unfashioable - Corba a good example.

There was a time when a Corba whizz was top dollar. Who would want to be a Corba whizz now?

I would suggest that if you want to be a top developer / architect you need between 4 - 6 languages and be able to switch between them.

httpete

Tim Robbins wrote: »

Even if you wanted to use Java on your back end you'd be mad to use EJBs. Use Spring instead and write your own framework.

EJB is problematic because they you are tied to your App Server. It's a 2002 approach. JSF is out because the fat client using JavaScript is the way to go now.

There are very few companies in Dublin with a good technology stack. Most IT work is working with legacy stuff.

I've started a new thread about regarding the presentation layer for Java EE web apps as I am writing one at the moment - http://www.boards.ie/vbulletin/showthread.php?p=82469910#post82469910

Regarding EJBs, I'm not sure what you mean when you say they are tied to your App Server.

1. Isn't the EJB spec the same no matter what App Server you use, and hence your code will be the same?
2. And even if not, how often would you be migrating an app to a different App Server in any case, I reckon it would be a very rare occurrence?
3. Is that the only reason to favour Spring over EJBs, as EJBs seems very simple to use, just annote the bean with @Stateless for example and inject it with @EJB.

(Note: If this question seems trivial it's because I am a PHP and Java SE programmer and I am writing my first Java EE application.)

conor.hogan.2

Anyone who calls Java dead I would ignore.