Slashdot Mirror
What Makes a Programming Language Successful?
danielstoner writes "The article '13 reasons why Ruby, Python and the gang will push Java to die... of old age' makes an interesting analysis of the programming languages battling for a place in programmers' minds. What really makes a language popular? What really makes a language 'good'? What is success for a programming language? Can we say COBOL is a successful language? What about Ruby, Python, etc?"
Portability and scalability are what win it for me, I like to write my code once and it's got to be powerful enough to deliver a complex solution.
Procrastinators, Unite Tomorrow!!
Java's well organized, has a great standard library and is (mostly) consistent with itself. Its only problems, as far as I can see, was that it was initially slow and that it marketed itself as a web language, when there were better choices for that.
Disclaimer: I've only coded in Java since 1.5.
Power: What can it do?
Performance: How fast can it do it?
Ease of Development: How fast can quality code be turned out by regular programmers?
Most modern languages fail on a couple of these. C is first class in Power and Performance, but it's not Easy. Ruby is okay in Power, and its very Easy, but it's slow. Java is Powerful, but doesn't match C for Performance, and it's not the quickest for development.
I'm sure many fanboys will disagree with my analysis. They'll say "Regular programmers don't matter (C)" or "It's NOT SLOW (Ruby)" or "Development is too quick! (Java)".
Really though, that's what it comes down to. The problem is, that there are unfortunate tradeoffs that have to be made. Most languages have a strength, but they all make sacrifices to be strong.
ad logicam Claiming a proposition is false because it was presented as the conclusion of a fallacious argument.
I think many people fail to recognize that the average age of software engineers has gotten higher and that many have realized that most of the pitfalls in software development have little to do with the language chosen. I would rather concentrate on good engineering practices and refining familiar modules I have developed than learn a new language.
love is just extroverted narcissism
Not to sound too much like Obi Wan, but many of the truths we cling to depend a great deal on our own point of view and all that.
If I was working for O'Reilly, Manning, APress, Wiley, et al I'd say a successful programming language was one which sold lots of books.
If I was a hiring manager for a large software company, I'd look closely at what language allowed the most cheap new grads to work together an produce something resembling quality code.
If I was teaching intro to computer science, I'd worry about what was preparing my students for the rest of their education.
If I was teaching a certificate-level course to people looking to get into the job market quickly, I'd look for the language with the highest placement rate.
If I was a person of little clue, I'd go largely by the hype. Some would go with the mainstream hype, and some go with the counter cultural "that's the big hype, but our language is better" underdog hype.
As a programmer, I prefer the language that helps me turn customer requirements into working programs that fastest with the least fuss on my part, and allows decent maintenance and customization later.
As the owner of a small boutique programming shop, I want my expressive, powerful language to give me an advantage over others using less expressive languages. I'd like to find others who can use it, but a few is alright as I don't need a huge team to work on programs.
Every program on your screen and your OS was written in C/C++
I can think of two ways to judge the success of a language.
1) If people use it.
2) If you can find people who will pay you to use it. And assorted corollaries: if people will hire you because you know how to use it, etc...
Given that programmers need to eat, I'd tend to go with the second though the two are basically related anyways.
Maybe it just means people using Python and Ruby have enough confidence in their code (rightly or wrongly) that they don't feel they should pay someone to review it.
Or, if you make the above information publicly known that you haven't reviewed Ruby or Python code that's a pretty good reason for someone to choose somebody else to review their code. Why would they pick a guy with little experience in that language?
I just started at a new job at the beginning of this year after quitting from my last job where I barely got to do any programming. The place where I work now is a Java shop. I was getting back to Java programming after a hiatus of a few years. For the last few years I mostly doing Perl with a smattering of C (PHP and Javascript on occasion). My experience with Java was mainly from college and a few odd projects I did here and there. The language had changed quite a bit over the last few years and to be honest, I surprised myself by being happy to get back to it (I had some sort of vague dislike for it for a period of time).
The company sponsored a trip to JavaOne at San Francisco earlier this month, for the Dev Team. I also got to go. This was my first time at JavaOne. It was amazing, exciting, and I learnt a LOT of new stuff. The main thing I got from there was that Java, far from being a programming language, is also a platform. There are a lot of new things being built on TOP of Java. For example, Groovy, and JavaFX. Java now has excellent support and frameworks to roll your OWN domain-specific languages.
Python and Ruby are not going to push Java out of the way. For example, you have mergers of Java with these languages (Jython and JRuby). Essentially you have Python and Ruby using Java resources and libraries. I think instead of "dying", Java is just going to evolve into a stable platform that lets you build stuff on top of it.
Vivin Suresh Paliath
http://vivin.net
I like
What makes a programming language successful?
Same thing that makes a religion successful. Adherents.
Lately democracy seems to be based on the skybox, the Happy Meal box, the X-box, and the idiot box.
You didn't review any C either, yet we all know that the language is out there and being used. Same with perl.
I think your field of work is too narrow to be completely explanatory.
Btw, I do agree with your general point - I don't see python or ruby bumping aside java. But your personal experience, extensive as it appears, is not enough to derive that conclusion
-Jeff
P.S. I really wish java would go. I hate the upper/lower case thing in all the names.
Please learn the difference between a dissenting opinion and a troll before you moderate.
Comment removed based on user account deletion
TFA:
"Flag on the moon. How did it get there?"
From a business perspective there is something to be said for older languages.
1. They have highly experienced developers that are widely available.
2. Apps written in them are generally older and have been time-tested and are reliable.
3. The language and its behavior is well understood and is well honed.
4. Many libraries are available
Changing to the latest and greatest language demands that you have a damn good reason. Hopefully you just have to port an existing app but you'll have to start all over with QA testing, security analysis, etc. Usually the reasons for changing are:
1. The pool of developers for that language has gotten too small to continue development and/or maintenance.
2. Hardware changes demand it.
3. The performance gain from changing is too big to ignore.
4. Compatability reasons with other apps, major customers, etc.
Making a programmer feel cutting edge isn't a good reason.
It is by the juice of the coffee bean that thoughts acquire speed, the teeth acquire stains. The stains become a warning
yeah, you know, 'cause when you have 50 programmers on a project, C l33tnesses like
while (x-->0) { blah; }
are so cool and easy to understand. and malloc()s make memory management so easy and cross-platform. and clustering is for wussies, if you need more than a core2duo on Linux, is because you're unl33t or because you need to do some routines in über-ELITE assembler.
now when you program in Java you forget all that crap, you just code. need a bigger app? J2EE it and run it on a cluster. add nodes a needed to keep performance. node dies? no problem, J2EE takes care of it.
migrated from mysql to Oracle or DB2? no problem, just let Hibernate know about it.
tired of Windows Server and want to run opensolaris, linux or OS X Server? no problem, just drop your EAR/WAR on the new server and relax. it's working.
wanna add more coders to your project? point 'em to the javadoc and let they read through the verbose (and thus self-explaining) code.
strong typing is there to keep you from doing stupid things. you can always tell what the program IS going to do in all situations, because you HAVE to specify all situations.
but you're too cool for java. lemme know when banks switch their systems to LAMP and we'll talk.
Do you ever think that maybe your survey has a heavy self-selection bias? I mean it seems to me that the most likely candidates for security reviews would be applications that have been around long enough to have somebody in management say, "Hey, we need to have a third party review this!". This explains how FIVE PERCENT of your applications are COBOL while only "three" are PHP. By your analysis, it's as if C/C++ doesn't even exist...
Java isn't going to die any more than C. Nor will Python or Ruby die any time in the foreseeable future.
Anyone can play Devil's Advocate and make one language look better than another from some point of view, but the fact is, different languages have their different pluses and minuses. I'm sure Ruby and Python have their pluses, but I don't see them being used NEARLY as much as Java. And take into consideration that Ruby has been around just as long as Java and Python has 4 years on both languages. If they were going to kick Java's ass, it would have happened by now.
I suspect the article is wishful thinking (though I can't read it 'cause the site didn't survive this post). I don't know why people have to make such a big deal about this stuff anyway. Languages evolve and new languages and paradigms will be created in the future. Computer programming is still in its infancy. There's a good possibility that 20-30 years down the road, none of these languages will be around. They may be completely replaced by some far more powerful paradigm we can't even imagine yet.
These kinds of predictions are old and pointless.
What's wrong with Java? Sure I can't slap together a web 2.0 site in 1 day like I could with .net 3.0 or Ruby, but they can't enable a high availability transactional based middle ware system. Java has so many great uses beyond simple web apps, it will always have a place in the enterprise and mobile devices.
You may find my appearance and demeanor foolish, but it is you who plays the fool.
> And repeat smart things like not treating arrays as first-class entities?
> Honestly, C is full of design errors.
Come back when you know how the computer works, grasshopper. C doesn't treat arrays as "objects" because the computer doesn't do that. If you want higher level abstractions, use C++, where you have the nice vector class that does what you want.
Just as an aside, you can use "EasyToReadAndEasyToType" in a language which is not case sensitive. Nothing is stopping you.
The real question is, should "EasyToReadAndEasyToType" be a different symbol from "easyToReadAndEasyToType"?
I happen to think the answer is "yes, it should be a different symbol". But that's a separate issue from whether you can use camel case in case-insensitive languages.
From the initial take, the intent of the article could easily be construed to say Java will die. But I thought was interesting that it would be from old age, not from some challenger language.
In many regards, concerning its lifetime, C has been/will be around for a very long time too... Which, I think, is probably one of the signs of a good language. Because if it was so horrible/people could not get their stuff done reliably, no one would use it.
It's not about the platform, language or the framework that makes an application safe, it's the security engineering that does. If you don't do any, your app WILL be insecure by design and there's no way you can't fix such code.
.NET (C# and VB.NET) since .NET 1.0
However, you have a point to a degree - I am initially more productive reviewing frameworks I am familiar with. But that doesn't mean I would be ineffective at reviewing Python or Ruby. It would take me about half a day to spin up in any language or framework as I found things that are missing. And that's the important thing:
I hate reviewing apps with zero security engineering. It's exactly like shooting fish in a barrel, but hopeless as you're not going to get a nice fish stew at the end.
What I look for are meta-issues found in all languages and frameworks. Syntax and functions can be found in online references - if you need them.
There is nothing special about any language as few protect against the security artifacts we look for.
For example, if your code has an access control mechanism, I look at it in situ on a live test app, deciding how best I might attack it, and then research using the code how I can obviate it at different levels:
* Coarse grained - is this feature access controlled at all? This is definitely a problem for J2EE apps that use servlets as folks think presentation level security is adequate. It's not
* Medium grained - does this feature offer different levels of access based upon your role? If so, how does this mechanism work? What do I do to get around it and steal stuff?
* Fine grained - does this feature restrict access to secured resources (direct object references)? If so, how does this mechanism work?
Each of the things we look at are verifying security mechanisms. Knowledge of the language or framework is simply not necessary. If you know what you're doing, you can prove the lack of security engineering by testing the app in situ and then research why it fails. Once I find a weakness, I look at the code to see why the weakness exists. Once I've found the issue, I look further afield for the pattern and then I document the issue. Rarely does an app or framework have just one weakness - they are usually patterns.
Picking up a new language or grammar and framework, like going from Struts to Spring MVC takes about half a day for someone like me who knows multiple languages, both functional like Haskell, or OO languages like Smalltalk or Ada, or scripting dynamic languages like PHP, Ruby or Python, or declarative languages like C or Java. We do not write the app, we are reviewing the app.
Security mechanisms are usually fairly clear if they exist. If they do not make themselves immediately obvious, they are usually missing.
Folks who have the hubris to think their code is somehow safe, like the COBOL folks on the mainframe or your example of not reviewing code if you don't know it well. That's why I turned down the Haskell review as I didn't know it well enough in the time available. If it was a longer review, I would have taken it as I love to learn new languages.
However, fyi, if you paid me to be a developer, I could be immediately productive in the following languages:
J2EE - Since Java was first released. Major frameworks include Struts, type 1 JSP with JSTL, Spring MVC, Struts 2.0, and JSF
PHP - Since PHP 3
Could code if absolutely required:
COBOL - 12 months review only experience
RPG - 12 months review only experience
Perl - 15 years experience
Shell scripts - 15 years experience
Ruby with RoR - tested it out for a new version of my forum (UltimaBB/XMB) but it was too slow
C - since 1985. Co-wrote the Matrox millennium driver for XFree86 back in the day
C++ - since CFront was a bastard child
Ada - since 1990. Still have fond memories
Pascal - since 1985, haven't used it for a while
Languages that I don't suck at but wouldn't claim any particular skills:
Andrew van der Stock
Bertrand Meyer
Object Oriented Software Construction, 2nd Ed., p. 881
Richard Stallman
GNU Coding Standards
Rob Pike
Notes on Programming in C
Linux Torvalds
Linux Kernel Coding Style
Patrick Lynch and Sarah Horton
Yale C/AIM Web Style Guide
class extended virtual theGreatEarthNovel::theProblemWithProgramming::theProblemWithProgrammingLanguages::theProblemWithJava tpwj{
...I really don't get it when people moan about java.
import inside here now system::ontology::basic::directions::fromAndTo::justToTestYourTyping::andYourNerves::To to;
towards::hardware::ui::screen::virtual::notMobile::notCLI where;
try::reallyHard::butItIsNP::4.times::atLeast:
system::interface::ui::utf::orLess::characters::ifFail(opticalRecognition) message;
message = "Java has no problems";
system::magic::lostMyYouth::expele::towards(to, where, message);
else:
system::core::shit::happens::throw::towards::withMessage("Whoops!");
}
"Reason number 2: Too much noise is distracting. Programmers are busy and learning 10 languages to the level where they can evaluate them and make an educated decision is too much effort."
okay, if you can't bother to learn more than 1 language, then you can't really call yourself a programmer.
If you can read this, I forgot to post anonymously.
I'll take any language that can let me write, read, and understand as fast as the speed of computers is progressing, i.e., exponentially.
I don't give a crap if language xxxxxxx is more efficient, more hardcore, etc. You know why?
Because I don't want to spend a year writing an application in C for efficiency and find out at the end that for a mere $1,000 I could have written the same thing in Python in a month and just bought a faster computer 11 months later.
YOUR time is linear, while the computer's is exponential. You'd be a fool to not take advantage of that and, frankly, type safety, efficiency, platform independence, programming style, power, etc. etc. can all go to hell. Just give me a beautiful language.
I am completely confused as to how the author can even ask the question "Is COBOL a success?"
Is COBOL old? Certainly.
Is COBOL outdated? Yes.
Has COBOL since been replaced by better languages? Yep.
Would you be insane to start a new, large, application from scratch using COBOL? Of course.
But "Is COBOL a success?" Without doubt, yes. Countless millions (perhaps) billions of lines of production COBOL code are still in use. It is still the core behind many of the applications that run our day-to-day lives. These applications have been running for decades with downtime records that would put an average "Web 2.0" app to shame.
Certainly, IBM deserves a lot of credit for this, maintaining pure 100% backward compatibility for those apps for the last forty years or so, but some credit is due to the language itself.
SirWired
Take *that* you young whippersnappers! If you want to program machines, you have to learn to think like 'em.
"My country, right or wrong; if right, to be kept right; and if wrong, to be set right." --Senator Carl Schurz (1872)
It's not a very good article. For one thing, measuring language use by search engine hits is silly. Just because it's the easiest way to get a number doesn't mean anything.
The current situation in programming languages isn't very good. It's been 52 years since Backus invented FORTRAN, and we still don't have it right. Let's look at what we've got.
So that's where we are. No really good hard-compiled language in wide use, and the major scripting languages each have some tragic flaw.
Ok, I'm going to chime in on this. What makes a good programming language is Strong Typing, Bounds Checking, strong compiler time checks and easy maintainability. One of my favorite languages was Ada, back in the 80's, because it made it very difficult to write bad code. I also used a version of GCC called EGCC which did the same sorts of checks and added run time checks, but it also never caught on. Coming from a Quality Assurance background, Ada satisfies the Maintainability, Portability and Reliability requirements. I was very happy to program in it after coming off Pascal and Modula-2. Marc
My company currently maintains two large python applications (40-50 LOC, not including custom support libraries). One of them is 6 years old, the other 2 years old. On our development team, we have familiarity with Java, C, C++, Perl, and the concensus is that we've had less maintenance work under python then the other languages.
If you add in the time spent on prototyping and testing, python has saved us way more time and effort.
Regarding type checking and reliability... You need to read up on the idea of "duck typing", Python's philosophy is that actual types get in the way, it's protocols and interfaces that matter. And after 7 years of python programming, I'd have to agree with that philosophy.
All of the critical parts of our apps have had unittests written (which catch semantic glitches at a level typechecking never will)... We've actually spent some time _removing_ type checking from the system, and replacing those lines with hasattr(obj, "append") calls, or creating synthethic protocol tests, allowing our implementations of various inputs to be widely varying: Jython implementations, CPython objects, who cares, they all LOOK the same.
OTHO, maybe you're trolling.
"Python looks like Basic" indeed.
Out of all the languages, I would have
never been reminded of that one.
Javascript maybe (especially considering what prototype.js has going on)
really... the reason Java is so popular is because all Universities use it to teach programming. The student are lazy and when they come out and play in reality they stick to what they know...
My favourite language for quick&dirty app making is Actionscript/flash. No other language comes close in speedyness... You have a vector based paint & sound editing program that can read photoshop files & import just about anything from avi movies to mp3's right next to your code, powered by a nice library... and your app can run in a web page or as standalone on practically all platforms.
So if someone is about to make a new language, check out the flash system and improve on it - Actionscript sure has it's illnesses.
In no particular order:
1) Development speed
2) High number of areas the language works well for
3) Low barrier to entry (ease to learn, expense and ease of setting up an initial dev environment to evaluate the language, community support...)
4) Available (cheap) Libraries
5) Industry Buzz
You could write your example just as easily like this:
l = [int(x.strip()) for x in l.strip().split(',')]
This version is arguably *easier* to read than your map-and lambda-based example. Map and filter are pretty much superseded by list and generator comprehensions (in fact, generator comprehensions allow for transparent lazy evaluation and are heavily used in good Python frameworks). reduce() is easy to provide on your own if you really need it, since functions are fist-class objects in Python, but even so I agree with Guido that explicit accumulation loops are easier to read than most reduce() expressions.
The only thing that I could conceivably miss are lambda functions. They do occassionally make for tighter and more readable code, but locally defined closures serve the same purpose, so again, there is no loss of expressiveness.
Define success? If we're talking about pure headcount then... I'd have to say ease of use and accessibility are the two only factors, because elegance, quality of implementation nor raw speed are factors in PHP nor classic ASP, both of which are/were very popular. All the glory seems to be in web development, people seem to forget that 95% of the developers at Microsoft are still writing C++ and rewrite 15yr old code bases to .NET isn't something MS is considering.
Your disclaimer is relevant, since you never got to experience most of Sun's big mistakes. Some of them will never go away, like the fact that some of the original library writers didn't understand the difference between byte streams and character streams. But all in all Java has a certain maturity it lacked when Sun was overselling it as the solution to all our problems.
But that's not why Java is in no danger of dying. Like any other software technology, programming languages are subject to lockin, that combination of cultural habits and legacy installations that keeps software around long after it's. Languages seem to be more subject to lockin than other technologies: unless it's a total failure right out of the gate, it soon develops a culture that maintains it, and a big body of software that must be maintained, and is too expensive to rewrite in a more modern language.
Consider FORTRAN. It was the very first high-level language (1953) and is full of design mistakes that reflect a total ignorance of artificial linguistics and compiler design theory by its inventors. (Not the designers' fault, these fields hadn't been invented yet.) And yet it remains the language for heavy-duty numerical processing. A lot of hard science and engineering grad schools rebel when told to learn it, but learn it they do.
Ironic story: I was working for Sun's Java org in 1998, when a corporate reshuffle caused our group to be given responsibility for maintaining Sun's FORTRAN compiler. Everybody's response was: What? Why the heck do we even have a FORTRAN compiler? Answer, the people who buy high-end computers for science and engineering won't even look at your platform if it doesn't have a FORTRAN compiler.
So Java (and FORTAN) will live forever. And centuries from now, newbie programmers will be wondering about those weird classes that are used to handle standard input and output.
Since state is encapsulated atomically, the interpreter is assured that function calls cause no side-effects,
This is an often repeated claim of Erlang fanatics. The truth is that, in Erlang, the functions themselves are the states. To say that there are no side effects in Erlang is a lie. Functions affect other functions (states) and if they are called in the wrong order, bad things will happen. Another problem with Erlang and functional programming in general, is that it does not support fine-grain parallelism. There are a lot of highly useful things that cannot be properly parallelized without fine-grain parallel processing, things like search and sort routines (I am still waiting for an effective parallel quick sort routine in Erlang). A third problem with Erlang is that, like multithreading, it is not deterministic. Determinism an essential requirement of reliable software. In addition, functional programming is not intuitive and many programmers find it hard to get used ot. So you people in Sweden and at Ericsson should stop promoting Erlang as the solution to the parallel programming problem. It is not.
Not every big library is bloated. It's only bloat if it has a poor size to functionality ratio.
For example libc is small, but it does not include XML parsing, HTTP support, SHA1 and MD5 sums, the ability to read compressed files etc. Sure there are libraries for that, but you have to pick and add them yourself. So libc is small not because it is amazingly efficient, but because it is limited in scope.
Personally, I like big standard libraries like Java and Python have. You pay for it in the initial install, but once that is in place, your application has access to a huge amount of functionality without having to add a lot of external dependencies.
No, I was saying Smalltalk is successful to me for some of the reasons I listed. I said it was not successful in terms of industry and popular terms. I don't think the 12+ language killers you are referring to are problems. Personally, those are likely the reasons I love the language.
Fileless language? Brilliant IMO and forces people to be organized in a structured way. VM? Great idea, makes deployment a snap, restore the state of your dev environment, etc. Need I remind you how VMs in general are all the rage now? Lack of extensive keywords? Makes it simple to learn and extend because the language is essentially written in itself, no weird dropping to C or other languages (see Ruby) causing all kinds of issues.
Outdated? It's updated more or less daily and even has several new web frameworks (just not 2.0 buzzword oriented like Ruby on Rails sorry, etc.).
Frankly, I find Smalltalk to be way ahead of the curve as far as keeping up with the rest of the pack in most areas because there are some great coders that share their stuff and churn it out quick, and the rest was already implemented in the past few decades anyway. How many times can we reinvent a datetime library? Lots of major changes != good language.
If you think no one ever used it, then you really lack an understanding of the language and its history. It's certainly more of a niche language than PHP, but it's not exactly an experimental research language either. I think you might be equating your own personal exposures to general usage.
In general, your first sentence sums it up. Smalltalk is successful because it implements both existing and new ideas well to the point where at least the ideas got more exposure, adoption, and notice. You can wrap all the features in a language you want, but the wrong blend makes it taste terrible. Smalltalk did quite a nice job and that's the reason people give it credit and other languages follow suit. Did you ever stop and think maybe there is a reason people give it credit? Note that I never claimed it was original as you pointed out.
As an aside, my favorite language for code reasons is actually Lisp, but I find developing in Smalltalk is much easier because of the philosophy and design on the language. I find Lisp more powerful, but less clear and about equally expressive for most things. I'd choose Lisp if I was going purely on language design, and Smalltalk for actual implementation from code to deployment. For me, Lisp tastes great but has a bit of an unwanted after taste in some areas. It's a personal thing.
Anyway, again, for me both Smalltalk and Lisp are successful because they help me do great work vs. other languages. I don't consider most popular languages successful because I'd rather die than write code in them, but unfortunately it's hard to escape reality when you're selling your services in the open market. I don't see how you can call Smalltalk a Sham unless you're trying to troll. All languages are good in some way, but we all have our preferences.
One does not learn their trade at a university. University prepares one for learning the trade.