Slashdot Mirror
Which Language To Learn?
LordStormes writes "I've been a Java/C++/PHP developer for about 6 years now. However, I'm seeing the jobs for these languages dry up, and Java in particular is worrisome with all the Oracle nonsense going on. I think it's time to pick up a new language or risk my skills fading into uselessness. I'm looking to do mostly Web-based back-end stuff. I've contemplated Perl, Python, Ruby, Erlang, Go, and several other languages, but I'll put it to you — what language makes the most sense now to get the jobs? I've deliberately omitted .NET — I have no desire to do the Microsoft languages."
I mean, I don't see php or C++ going anywhere anytime soon....
-- www.RoachMcKrackin.com
Still in demand and it will not die.
Industry constantly tells the Universities they need more C/C++ programmers for industrial systems. If all you are looking at is web based development, you are seriously limiting your options. I suggest a less restrictive filter on your search parameters.
Every mans' island needs an ocean; choose your ocean carefully.
80% or so of all web-backend postings I see are PHP/Java/.NET or the like. The other 20% are all Python (usually Django, though I prefer Pylons myself) and Ruby. If you want to pick up another language just so you can be future-proof, go with Ruby. I haven't learned it yet (I do javascript myself, and use PHP or Python when I do backend), it seems to be a more common request than any of the others you listed.
I am become
"What language makes the most sense now to get the jobs?" What jobs?
.NET development is taking off whether you like it or not. With oracle serving up a shitstorm over Java it's only going to gain more traction. Omitting an entire language and framework simply because it's developed by Microsoft is a pretty poor reason especially when it's gaining use in the very type of work you're looking for (web-based back end stuff). Honestly if you're looking for a job, consider learning C# and familiarizing yourself with the .NET framework
Have you thought about one of the languages spoken on the Indian sub-continent?
Nate
I would go for Chinese.
I've deliberately omitted .NET — I have no desire to do the Microsoft languages.
Poster obviously has no desire to be employed either. Love it or Hate it C# is pretty much the only langauge in demand by big business these days in the UK unless he's perfectly happy doing small freelance jobs etc which PHP is fine. Other languages he's mentioned are all pretty much unused apart from in the domain of nerds but certainly not by the majority of the companies recruiting. Ironically enough I reskilled from C# and other .NET oddities to PHP a few years ago purely out of personal preference.
Last I checked, being able to create apps with native hooks on the Mac platform is the hottest shit steaming right now.
Win or lose, either way you'll earn money.
No language is more universal. No language is more direct. It will never die. It transcends trends. It is the only decent language to me, having tried way too damn many in my life and always left wanting until I return to C.
It is the perfect language. People might gripe that it's somehow "obsolete" or missing "modern" features, but to me, that's part of its appeal -- you get to do with it exactly what you need to do, and that is the essence of programming to me. Leaving too much to the language makes me feel powerless and less in control.
I love C. If it was legal, I'd marry it.
You need to drop the Microsoft hate if you actually want to be employable.
I think it's typically assumed that if you know how to program, you ought to be able to interact with a standard relational database. There's almost no prospects out there for someone who does SQL and nothing else...
Then you have just limited your career. But don't let me stop you, the rest of us want jobs too.
---- Booth was a patriot ----
If you knew .NET I'd have a job for you right now. Love it or hate it, MSSQL is still the fastest kid on the block, and its .NET reporting tools aren't going anywhere anytime soon.
You should learn Chinese.
I intentionally tried to avoid MFC, and learned it anyway. I avoided .NET like the plague, and work moved me right back to the plague.
Since .NET 2.0 it's been a stable API, and if you're going to do web or web/desktop development, it's a good thing to have in your back pocket. And I'm saying this as someone who intentionally avoided it.
I picked up Prosise's MFC book so I'd know what MFC was doing behind my back, and I dropped wxWidgets once it became clear it was an MFC "port" - if you don't believe me read the wx history. I intended to stay classic MFC all the way, and learn something else - anything else (but Java, that's my ideology and just as unfounded). Qt and... whatever the dominant web language was in 2001.
I write .NET for a living. If nothing else, you can be read-only with .NET like I am with Java. I can search for an algorithm and find a public domain or otherwise compatible implementation, and if it's Java I can port it in a few minutes and have what I need - whether it's .NET or C/C++, which is where I prefer to work.
Learn .NET, even if you are working in a full open source shop. There are lots of open source programs available only in .NET, and a free compiler (not the GUI, just command-line).
I don't have mod points, so I'm just backing up dreamchaser (49529). I can write x86 assembly (att or intel), C (K&R, C89, C99), C++, VB5/66, VB.NET/C#, ASP 3, JavaScript, VBScript (cscript and IE), SQL (MS and Oracle) and lots of others less proficiently... so it's not like you can't learn multiple languages. In fact, the more you know the better. I write better .NET code because I think in assembly when performance matters. I write better ASM code because I think in OOP when code clarity matters. Yes, I probably need mental help, but the more you know the better you will be. The more ways you can think about something, the more solutions you can weigh when you have to actually implement something.
Here's the best part. Learn what .NET does *wrong* and avoid implementing that in your apps, or avoid using constructs like that in whatever language you get paid to use. Learning .NET has made me a better C++ programmer, far more than any other experience in my life. Both for the good parts and the parts that could be better.
You'll want to learn to use ILdasm if you go this route, no question. Obviously my vote is .NET.
Search sourceforge for stuff in .NET languages, C# is probably going to be more familiar, download the free compiler from MS, compile, make changes, and start reading.
Seriously. My life is made hell by one stupid microsoft idiocy after another day after day (I manage a server farm of Microsoft VMs). The fact that they treat their development community like crap (Classic VB, f'rinstance) and abandon products with... abandon doesn't help much.
Despite this, that's where the jobs are and all the crabbing of myself and the development community hasn't changed that. I hated MS in 2000. I hate it in 2010. I expect to hate it in 2020. And it's not going anywhere. Profits are up. Like the air, it exists. And I'll still be cranking out C#, ASP.net, or VB.net or whatever is called for.
Please do not read this sig. Thank you.
+1.
Don't learn the languages. Learn the the paradigms.
Once you know a paradigm, picking up a new language under that paradigm will be just "yet another language", and you can learn one in a week (or 7 in 7 weeks). Of course, it will take more time to actually become fluent in language specific idioms, standard libraries etc, but those are not rocket science either.
If you know Java, C++, and PHP for 6 years, and cannot pick up any arbitrary language in a weekend, then you should find another business immediately.
You've clearly failed to grasp the concepts of programming.
I've always been puzzled about the focus on languages. A language is just a tool or a palette. The important thing is being able to design software in a clear and efficient manner. Languages can typically be picked up in a very short period of time. I understand that the people who only understand buzzwords are blindly seeking a particular match, but if they were smart, they would look instead for a person who was a skilled designer of software, rather than a person who knew a particular language.
In my career so far (I'm sure that many others have a much broader list), I've had to write software in: assembler (IBM mainframe, 6052, 8085, Z-80, 8051, Z-8, 68K, X86), BASIC, FORTH, FORTRAN (4, 66, 77), RATFOR, lisp, C, C++, PASCAL, PROTEL, perl, tcl, python. Java I've written a couple of example programs in, but have never written in it professionally. But I presume it is no harder to become proficient at than all the others. ADA is a good language; people are starting to realize the value of the rigorous protections it applies. But yeah, we were writing ADA code back in 1995, and it didn't seem widespread outside the military at the time.
If you want to be another common fish in a huge ocean, learn C# and sharepoint development. If you want to be hip and cool, but are willing to compete with low price coders from developing countries, go with LAMP development. If you want to be a big fish in a small pond and can self promote and communicate well enough to pull it off, pick something painful but useful to corporations ( Rational / Websphere / Oracle / Siebel / SAP development ).
I do most of my client based work using Lotus Domino as a back end server and data platform. The development IDE is freaking horrible compared to visual studio or pure eclipse. The documentation is poor at best. There are a lot of workarounds you have to know. In many respects, it's a terrible thing to have to learn. HOWEVER, I've been doing it for a long time and am very very good at it. I'm never short of work, and I can accomplish things with it for my clients in less time and at less cost than any other platform I've ever found. I also use visual studio to build desktop applications, c++ to write custom modules for my Asterisk servers, javascript for web front end stuff, bash shell scripts for linux back end stuff, etc etc etc.... Right tool for the job and all that.
I know by writing that I'll draw a bunch of crap from cool kids that hate the platform I use to make a living, but I'm willing to bet most of them would trade annual incomes with me in a heartbeat if I gave them the chance. I've managed to have my own business for close to 18 years by focusing on what works rather than what's cool -- and by never letting myself be just another commodity programmer among a giant pool of people with similar (and frankly better) skills.
The problem with quotes on the internet, is that nobody bothers to check their veracity. -- Abraham Lincoln
Clearly you didnt learn Perl, otherwise you wouldnt be calling it PERL.
I'm going to assume that you know perl is not considered an acronym (anymore at least) and that you're just using CAPSLOCK as CRUISE CONTROL FOR COOL.
This is what happens when I omit the tags. Sorry. No, I'm not interested in Miguel de Icaza's attempts to get Microsoft's patent-encumbered technologies into Linux - always one generation behind. That's not cross platform, it's just porting the chew-your-leg off environment to Linux so you can have that fun experience over there too. I'm starting to believe that stupid is contagious. No wonder why he doesn't want to learn the Microsoft stack. He doesn't want to catch it.
Help stamp out iliturcy.
Tying one's career to careful thinking is always smart. Do you really want people easily de-compiling your code? Microsoft is the British Petroleum of software. Eventually there will be impossible problems.
Look, I'm not fan of .net (or the company) either - but dude, it's been around for a decade now; usage isn't declinging and just like any actively developed product every couple of years brings new improvements. It's great to say th ere will be "impossible problems" - but if we're realistic, we see they will be no more impossible than any others. As far as de-compilation: wait, I thought information wanted to be free. Yeah, I know that was a straw-man - it just slipped in. Anyway... you could say the same about any language that compiles to bytecode -- singling out Microsoft because you don't like their practices is just silly. (It's no harder or easier to decompile .net than java -- though interestingly the tools to decompile .net are more mature, it's true -- stilll obfuscation goes a long way.)
A full, complete version of Microsoft's operating system, Windows 7, costs $300, about half the cost of some laptops.
And an OEM version - readily available - costs < 100, so what's your point again? That some of the tools we use cost money? And this is a reason not to use them?
Eventually Microsoft's abusiveness will cause an Enron-style breakdown, in my opinion.
First BP, now Enron. Dude, really? I suppose I should be thankful you didn't throw a couple of "M$"s in there. Perhaps you could next detail a breakdown of what or the type of impossible problems we should be expecting. Until then, you're just a troll. And one that I'm feeding, at that.
Your point being? If you're billing your time at $150/hr, that is 2 hours of work.
People harp on about the cost of Windows, etc but in the real world it is often irrelevant if it saves a couple of hours of time.
I run: Windows, OS X, Linux, FreeBSD. Just because you have a hammer, doesn't mean everything is a nail.
Most of Microsoft's own software is NOT written in .NET. There is a reason for that.
Is it because most of Microsoft's own software was written before .NET was released?
I don't care if it's 90,000 hectares. That lake was not my doing.
If only whitespace made as much sense to humans as as it does to computers... Haskell and Python might actually be good languages.
Where I work, I currently do one interview a week. I only said "hire" twice in the last year or so. Truth is, 95% of people I have interviewed so far couldn't write decent code on the whiteboard if their life depended on it, in _any_ language. Your fear is misdirected. No decent employer gives a shit about languages in a job interview. They care about whether you can write the fucking code, in the laguage of your choosing, and whether you have experience in the areas you're applying for. I.e. if you bill yourself as a backend dude, they'll want to see if you know e.g. distributed systems, and have the backend mindset. If you're a frontend guy, that's another set of skills entirely, but still very little (if anything) depends on the language. You can learn the syntax in two days. You can learn the libraries and language-specific idioms / patterns in 2-3 months (if you're proficient in at least a couple other languages). It's not that hard.
And if the employer makes the assumption right away that you _can't_ learn e.g. Ruby on Rails, to hell with them. You wouldn't like working there anyway.
"but you'll need to learn python"
What's to learn? You can teach yourself to be functional in it in no time.
Languages have gotten so boring, these days. I know, the languages are the same, it's just the fact that we've learned all the new concepts and the only thing really novel in most of these languages is syntax. Know some lisp, some fortran, some pascal, some java, some groovy, some ruby, and what's left? Anything new under the sun? Any real new concepts? (Not just ranting here -- anyone with some suggestions?)
We should be telling this guy to learn ARM assembler. At least that'll challenge him.
The thing this recession taught me more than anything is that corporations have no morals, no ethics, and really, no just claim to fair treatment. They are not humans, and not deserving of anything more than that for which they pay. Not a dime.
Corporations - particularly large, publicly traded ones - routinely use "the recession" as an excuse to treat their employees like dirt. Get off your high horse - bowing down to your corporate masters so you can "have a job" only screws yourself and your fellow employees. Do us all a favor and stop working - or at least demand the respect that you, a human being, deserve.
The fact that the economy is in the toilet doesn't change the fact that you're a human being and deserving of the respect due a human being. If you think otherwise, well, you're just as much a part of the problem as the companies which exploit the poor economic situation.
The society for a thought-free internet welcomes you.
Lisp already exists.
If Lisp did not exist, it would be necessary to invent it.
ID: the nose did not occur naturally, how would we wear glasses otherwise? (apologies to Voltaire)
The language I enjoyed the most hands down has to be Perl, the community was awsome. (Hit a few YAPC events to meet them). london.pm the irc channel is almost always helpful now days (as aposed to the sarcastic replies you used to get).
Perl/Catalyst for the win.
> Once you know a paradigm, picking up a new language under that paradigm will
> be just "yet another language", and you can learn one in a week (or 7 in 7
> weeks). Of course, it will take more time to actually become fluent in language
> specific idioms, standard libraries etc, but those are not rocket science either.
I know people who take the same approach to natural language. After all, Spanish and Italian are very very similar, aren't they? The reality with natural languages is that "all languages are the same" thinking enables you to abuse several cultures without actually understanding any of them.
And I think that to a large extent the same thing goes for programming languages. For example, if one of your "paradigms" is "object-oriented", does learning Smalltalk really prepare you for making best use of OO in Java or C++? Or vice versa? The inventor of Smalltalk and OO certainly doesn't think so.
I spent some time a while back trying to explain Scala to a Java programmer. His response was "It's just like Java." Well, Scala *is* just like Java, as long as you ignore the huge and central features that are not like Java. When I started to show him those features, generally in a "replace a page of code with one line" sense, his response was "I don't like it", and that was the end of the conversation. That, in practice, is what "learn 7 languages in 7 weeks" looks like.
My defining experience in this context was observing a government contractor whose preferred language was FORTRAN, who was told he had to code in Lisp. I would not previously have believed that it was possible to write Lisp as if it was FORTRAN, but that contractor proved me wrong. And, to be fair, I find that I have to make a conscious effort not to write C++ as if it is Lisp, eg "everything on the stack and screw the efficiency".
"7 languages in 7 weeks" only works if you stick to programming with the features that can be found or kludged in just about every language. Nowadays that's going to mean procedural code with loads of variables and a bit of OO for accessing libraries. It works, but it's a recipe for terrible, terrible code. But, hey, it will be equally terrible in 7 different languages!
Virtually serving coffee
I think it's typically assumed that if you know how to program, you ought to be able to interact with a standard relational database. There's almost no prospects out there for someone who does SQL and nothing else...
They actually have titles and everything. They are called DBA around my shop. The good ones are like old-school unix guys who wax poetic about their favorite shell script and kernel optimizations.
If you are a PHP or other front end developer who creates SQL to power it, you are very, very likely not a SQL master. A good programmer with experience can create SQL databases and queries that work well. His code will be amateurish and inefficient to a good DBA. They do the same thing OS programmers do, delving into the deep inner-workings of the database engine to find all the little tricks, optimizations and security gotchas.
Good DBAs tend to be more math oriented personalities than the larger developer population. Probably because they have to live in a world dominated by set theory and complex logic.
BTW, if your experience with DBAs is a bunch of Microsoft Certified Professionals who are proud that they can create a stored procedure to fill a ticket - then you haven't been working with a good DBA. Those guys are the equivalent of the "web developer" who can use the GUI development environment to put a couple of forms together. A good DBA will take that query that you spent two days optimizing to get from 15 minute run times to 2 minute run times and get your results in milliseconds. Often the optimizations they make won't even seem logical to the untrained - until you watch how much faster they run. They are able to do this because they've spent years focusing on one platform.
Still don't buy it? Ok, a quick example. One of my analysts was faced with a set of tasks that was taking too long and causing application timeouts. These tasks involved importing and parsing millions of rows and then joining to many tables of tens or hundreds of millions of rows in a highly transactional environment. After banging his head against the limitations of the database engine for a week or so, he finally decided that he needed to expand the functionality of the engine. So he added a couple of customization DLL's to the engine (written in C#) to add two new commands with the features he needed. He was able to get an already well-optimized run time of two minutes down to about 35 milliseconds. Oh, and my team is already finding lots of other places to use the new features he added, knocking a few percent off of the CPU load on the server and improving response times.
Pick a direction.
If you like web dev, don't look at a language, look at a path. If you want to stay relevant in the front-end coding, Javascript, getting good with HTML5, and one or more of the various things that feeds both (Python, Ruby) is good. Do it via learning some widely used package of functionality. Write something missing, and release it (Django tagging system, with a Canvas tag cloud?) Having released code makes you much more employable. On the back end, well, Python or Ruby + knowing the oddities of one of the NoSQL tools is nice. (And by "knowing", I don't mean read the O'Reilly book, I mean put them under pressure and see how they fail.)
If you want to become a better programmer, learn Lisp. Really, no shit. It changes how you look at things.
If you want to become a better programmer, but don't want to invest in Lisp, pick pick up a functional language, like Haskell or ML. But do Lisp first.
If you hate the idea of either of the above, get better with your current language of choice. Write an ORM, or a templating language. Nobody will care, but you'll learn why everyone hates but still uses the ones we have.
Learn Objective C, or the Android API, and write something for a phone. We're just past the "here's my todo list, here's my Tetris clone" phase, and it is new territory, and one of the few genuinely interesting things to come about since the mid-90s. If you show some initiative, you can land a safe spot doing this. By about '15, I expect most folks to have a smart phone and be sorted as to expectations, so that's about three years to get good at things and maybe do something interesting to set yourself apart, at the most.
If all you care about is shooting for the center of mass, learn either Microsoft or get good with PLSQL + the weird crap Oracle makes you do to bundle Java up for database deployment. I don't know what Microsofties do, but if you get good with server-side Oracle code, know your shit with performance issues, memorize various oerr codes, and can parrot back whatever whitepapers Oracle released recently about X And The Enterprise (you know, Private Enterprise Cloud Computing, or whatever), you're very employable. Bonus if you've lived a clean enough life to get a security clearance. Check your ego with the receptionist and pee in this cup, please.
I forget what 8 was for.
I'm puzzled by what you mean by "drying up". I'm constantly seeing Java and PHP jobs. Granted I'm in a tech hub (Austin), but there are tons of unfilled jobs. The fact that Android runs a java variant puts Java even more in my demand. I'd suggest adding Objective-C and Javascript and you'd have pretty much every current desirable tech skill.
Have you considered that the problem isn't your technological skills, but where you live?
You cringe? Interesting... It is ONE char and only one.
When I am coding I there are times when I just drop a quick test into the left hand margin because it is visually a "sore thumb" that sticks out and is unmistakable as something that should not be there and if it IS there then it requires immediate attention.
While Python is a very nice C Clone they really fucked it up with this bit of stupidity.
Hey KID! Yeah you, get the fuck off my lawn!
...despite not strictly needing to!
Hit the nail right on the head my friend.
All the dogma about dynamic languages giving the programmer the freedom to build like their imagination wants to goes right down the drain when the language designer imposes their layout style on the people who would use the language.
I wonder how many lines of cruft could be removed from the parser if it did not have to manage indentation?
Python is no more nor no less elegant then C since it is just a C clone. I like you have been doing this shit for too damn long but I gotta say I have never really had a problem returning to well written and well commented C code.
Hey KID! Yeah you, get the fuck off my lawn!
Agreed (from what I've read about Python). Perl is a bit more prevalent and has a longer history, meaning there may be more people/places using it and there's more code out there (not just CPAN).
From a functional point, if you know one you don't need the other, but if you don't know either, I'd suggest Perl. I know there's a little religious waw between the two, but it seems silly. That said, I think the white-space delimited block syntax in Python (and other languages) is really stupid. I know you Python people will chomp at the bit about that, but I'm right about this, Guido is a snob about this, and you know it - let it go. Anyone who's had their Makefile blown because of a lost tab, or bitched because X converted tabs to spaces in a copy/paste knows what I mean - and yes, get off my lawn :-)
It must have been something you assimilated. . . .