I can't speak for the commercial world, but in the academic/libraries/humanities IT world where I live, C# would be basically of zero interest to us.*
We've interviewed some people with C# backgrounds on the basis that if they seem to be decent, they can turn their hand to the things we do need/want, but it's not remotely on our radar as a desirable skill.
* we basically need Python, Ruby, a bit of Java, and Javascript. The last, in particular, we find really hard to find/recruit.
I'm dubious that there's much adoption of C# outside of the Microsoft ecosystem - it's technically possible, but I'd be surprised.
But they're very close languages and a lot of the skill will be transferable (APIs are a bit different but honestly any actual programming work will require learning reams of new APIs for the project anyway). C# is better than Java in a bunch of ways and I think it's valuable to know what kinds of things are possible even if your current tool doesn't support it.
I feel like I was on record as arguing against Java in the first place, so let me reiterate: you're going down a terrible path that will lead to you having shitty, soul-destroying jobs!
You could do a bootcamp, like the kids are into!
Javascript. The last, in particular, we find really hard to find/recruit.
...really? I'm astounded. We've moved a lot of things away from Java into Javascript both because we're going web native and it's easier to find talented Javascripters. But it might depend on what you're doing--we wanted UI developers, and a lot more people know JS/HTML/CSS than Swing.
"Mono makes C# non-Microsofty" is something I've literally been hearing since 2004. But yeah, everything I've seen has shown it to be very similar to Java. I suspect that you could learn one and easily transfer into the other, with only some slight annoyance at syntax difference and what the basic libraries have.
I admittedly am envious of the convenience operator C# got recently(ish?) that lets you follow a chain of references until you get a null, saving a lot of repetitive null checks. There are patterns that make this less useful, but you have to engineer them in.
There's a class of resumes I see of job applicants that I can summarize as "Microsoft salt mines", and I understand wanting to stay away from that. But there's another type, at least as common, that could be called "Java enterprise salt mines", and it doesn't seem any better, just less Microsoft-y.
That doesn't help much, except to note that being language-agile is probably the truly valuable skill, compared to knowing any particular language.
(Seen on Twitter recently: "Put Haskell on your resume even if you don't know it. When asked, say your resume is lazy and you'll learn it when results are needed")
Does that salt smell the same in either?
If you want to learn SAS and move to Arizona, I know a guy.
re: 4.1
May just be an indication of how difficult it is for us to recruit, in general, when the skill in question is in demand, and when people can work in London for much more money than we can offer. But javascript has been a perennial problem. For UI stuff, yeah.
As a general rule, at least in the world I work in, we don't really care about compiled languages at all, except for certain specialist tasks.* Interpreted plus JIT is basically all I ever work with.
* image processing and image delivery stuff, basically, but the code wrapped round the open source C++ applications is still all Python or Ruby or (erk) PHP.
I don't know what things are like there, but here the university seems to have decided that local costs of living are low enough and benefits are good enough that they can offer basically the same wages as ten years ago. This works reasonably well for somebody like me, because I bought a house ten years ago. It doesn't work nearly so well for somebody who needs to pay the actual current housing prices.
3.1.2 is the truest thing that has ever been posted here.
Learning java may be a sensitive indicator, but it isn't very specific.
Enh, I've been working in Java professionally for about eight years, in the same job. Which is probably indicative of a rut, but my work has greatly varied over the years: started out doing Swing UIs, but have mostly been wokring on what I think is known as "business logic"--our mildly AIish tutor framework, as well as support for the tools cognitive scientists use to write models in. In recent years we've been compiling to Javascript via GWT so I've been able to work at a lot of different levels.
I doubt my experience is typical. I don't really know Enterprise Java, and I'm only slightly interested in doing so. Seems like it's difficult in a way that isn't satisfying.
I do wish I was working in something more functional, like Haskell or OCaml or at least Scala, but I'm assuaging that by finally getting to use lambdas (we only recently upgraded to the version of GWT that plays well w/ Java 8).
1: My last workplace had some in-house C# applications. They were terrible on the user side, but that might not have been C#'s fault. I think they've now ported them to some combination of Java and Ruby.
Python and C (along with various *nix tools) should be sufficient to get a good job, right? At some point I should probably try to find one that pays decently, but job search fears and general CHANGEBAD, coupled with a strong desire to finish several large personal projects that dovetail nicely with my current job have made it extremely easy to put off. Of course, I can continue hiding from reality and trying to finish my projects indefinitely, and presumably at some point I'll be too old (if I'm not already).
compiling to Javascript
The enormity of the fact that this is sometimes (often!) a good idea in the modern world never dims.
compiling to Javascript via GWT
The scope creep of the global war on terror has gotten out of control. I blame the close relationship between Silicon Valley and Obama.
I don't even know what "compiling" is, but, according to my job description, I'm a programmer.
Our CTO tries to say that noise (re: C#) every once in a while because he worked on a .NET project many moons ago and fell in love with VisualStudio, but the rest of the engineering team regards it derision, rightly IMO.
I think both Java and C# are strongly correlated with salt-mining jobs. My last employer (named for evil prophetic rocks) was a Java shop and definitely a salt mine, and I needed to re-tool before landing my current gig.
Javascript stuff may be the fastest way to get from zero to employable, but there's so much turnover in tooling and best practices right now I imagine it would be really hard to figure out where to get started. I know in the past I've recommended Ruby because of its roots in our fair city, but honestly you may be best off picking up Python. Then in a year you can spend six weeks playing with Numpy and call yourself a 'data scientist' and get a 40% raise. Starting with Ruby and JS is more likely to lock you into just doing Web stuff.
I've heard great things about http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-00sc-introduction-to-computer-science-and-programming-spring-2011/ et seq., and it lists on a resume just as well as bootcamps or local in-person coursework.
I dunno, isn't LINQ supposed to be pretty neat?
Is the phrase "salt mining" common jargon among programmers? What does it mean? Google is not helping.
I don't even know what "compiling" is, but, according to my job description, I'm a programmer.
Compiling is when your LaTeX document doesn't work.
C# is definitely Microsoft-only. I interview a lot of applicants and I've only seen one resume with non-Microsoft C# experience (she was doing cross-platform mobile using Xamarin).
I personally don't care what language experience a candidate has - I generally recommend to hire if they can answer simple algorithm questions. Really simple stuff - "can you write a function to find the largest element in a list" sort of things. I'm shocked at how few people can do that. Its a minority, certainly (although possibly because of how stressful and unnatural an environment an interview is).
I only start to care about relevant experience when we're hiring at the principal level - and even then I don't care about specific languages so much as similar sorts of domains.
But I think our HR department, when doing the initial screening, does more of a keyword search on available resumes and you have to get past them first.
20: does it seem like it would be fun or creative or intellectually rewarding to work in a salt mine?
That really depends on the particular mine.
23: But on the plus side it wouldn't be smelly.
I've set simple problems for the last few (unsuccessful) rounds of developer recruitment. Where I ask, for example, how they'd sort metadata records or images for a foliated manuscript into the correct physical order.
I don't particularly care if they go into the details of sort algorithms, I kind of expect that people will just use the built in sort functions in whatever language they are using. But I want to see evidence that they can understand the problem, know how to approach it, how to get the relevant information they need from the sample data I've supplied, and so on, and that their approach will actually work.
Then I start chucking refinements to the problem that make it harder at them.
'What if there's a mix of arabic and roman numerals?'
etc
Mostly what I want is someone who can get on and work without micro-management, as I (and the other people who manage technical projects) don't have the time to stand over people telling them what to do.
Salt mine ..... I remember a senior computer scientist type once called 'em "the cobol dungeons". As in "Java/C# is the cobol dungeon of the 21st Century".
Spent 20yr working pretty near or in those dungeons. Will never do it again.
I second people saying Javascript. Also Python. And esp. those who push actually learning how to write real algorithms -- that skill never goes out of style, b/c it's closely correlated with being able to solve simple problems right the -first- time. Seriously, that's what Google's really interviewing for, with their insane write-an-algorithm-on-the-board-while-I-watch-you-sweat interviews: they're looking for people who are so good, they won't screw up the simple stuff. B/c seriously, as somebody else said, there are so many people. who screw up the simple stuff.
But back to what to learn: yeah, JS, Py, that MIT course has gotta be a good 'un (notwithstanding that I shed bitter tears for the eviction of Scheme). And don't worry about the fact that things are changing fast in JS land -- that merely means the barriers to entry are lower, than in other areas, where things don't change fast. Seriously.
But here's another 'un: Unless you're olllld, why not learn to write mobile apps? It's all the buzz, eh? (and if you're ollld, well, it'll just be harder). But for anybody in their 30s, it seems a pretty good bet, no?
Java is used more for "back-end" coding. Javascript is used for front end (web page) coding. Python is used to make Real Programmers' heads explode, but since people love to do that, it's popular. All three and C# have roughly similar syntax but are very different in actual usage. Back-end coding is more likely to be "less fun," I guess. I mostly do Java these days but it's research-oriented projects.
12. Haskell, for all its formal beauty, takes (well, to be fair, took me...) a long time get my head around it, even though at base it's got a lot in common with Lisp.
Ogged, learn Python or Javascript (bearing mind what LambentCactus said in 18), but if you do either it would be nice to learn from a source that doesn't do hack and slash coding (meaning "your successor maintaining it will hack you and slash you"). They are easy languages to misuse.
It wouldn't hurt to take a formal programming course or read the classic "Design Patterns" book, aka the Gamma book.
(ps: Didn't you ask about this some months ago as well?)
Pretty much pwned by 27, with whom I agree. "Introduction to Algorithms" is the go-to algorithms book, an expensive (and very useful) doorstop.
If you mind-meld with those two you can ace a Google interview!
Come on, Javascript is super heinous. Why not learn something elegant, like ... Purescript!
I think SAS is what's going to be the future.
Thanks, all! I did post about this a while back. The idea behind doing Java was that there was a determinate course of study, and since, as a dilettante, I worry about unknown unknowns, something approximating a curriculum with a test at the end was appealing. I figured it was a way to learn the basics, and then apply them in other languages. It was going very well, too, and I could fairly fluently crank out solutions to the toy problems in the book I'm using, but then we bought and renovated a house and I basically dropped it all for six months and forgot a lot, and now I want a course of study with a schedule and end date.
Bootcamp sounds nice, but as the primary caregiver to my kids, I can't swing the schedule. There's a bootcamp here that meets nights and weekends, but I think it's for Ruby? That would probably lead to employment but then I'd be doing solely web stuff, I think, and, again, I'm hoping for a broad base of learning.
I tried learning Python a few years ago, and loved it, but I think I was worried I wasn't mathy enough to be a convincing candidate for most Python jobs, and there was also the set curriculum "advantage" that Java had.
Seriously, that's what Google's really interviewing for, with their insane write-an-algorithm-on-the-board-while-I-watch-you-sweat interviews: they're looking for people who are so good, they won't screw up the simple stuff.
I call bullshit. You can find that shit out other, better ways; they're asking algorithm questions because they're the kind of thing you learn in school and then (unless you're doing certain kinds of work) forget about over time. Which is to say, they're a deniable proxy for age. (Also dick-waving.)
The guy who interviewed me on the phone for Google sure wasn't trying to find that out.
I was worried I wasn't mathy enough to be a convincing candidate for most Python jobs
This is an unsound fear.
37- He didn't ask you to send a picture of you waving your dick?
Novalis worked in the salt mines.
Honestly, writing algorithms (in the colloquial sense of an algorithm being a complicated but small bit of code, say the size of a function or two) isn't hard; putting together larger systems is hard. If you do that well, you can write/use the trivial stupid algorithm and substitute in a better one as you find you need to. The Design Patterns book is a step in the right direction towards doing this. For the most part it doesn't matter too much what language you choose to learn this in. As said I have a bias towards Java and things strongly typed in general, but it doesn't really matter; every language is frustrating to use in its own special way and you'll learn to deal with that.
||
Is the GOP debate happening now? I thought it was going to be 9pm EST.
|>
Also, I'm a believer in Turing completeness (even if HR resume screeners aren't). If you learn one language well because it has a structured course of study, it will be easy to pick up Python or Javascript on your own afterwards. If your timeline isn't 'must find programming job ASAP' go ahead and learn C# from this program and teach yourself Python after. It sounds like you may be more likely to be a happily employed programmer 12 months later using that strategy than jumping in to Python self-study. If that's the case, that's a totally valid path. I started out in C and haven't written a line of it in a decade.
The preliminary consolation prize one just ended. The main one with Trump is later.
Ah thanks. Don't think I'll be able to stay up for that.
It is still at 9PM Eastern, though. Do you have your time zones straight? Also, it's technically EDT.
My bootcamp uses a full-stack JavaScript curriculum for many of the reasons already stated. This is the smartest basic intro IMO: http://eloquentjavascript.net/
Also looking at the source of this library to see how it implements some of the canonical functional programming abstractions:
https://github.com/jashkenas/underscore/blob/master/underscore.js
If you are going enterprisy, learning C# when you already have Java basically doubles the amount of jobs you can apply for. I wish I knew C# - there are some nice gigs I haven't taken because I didn't have it.
Yeah, C# basically means Microsoft, but there are other niches. One is the Unity gaming engine. If you learned C# you could spend time fucking around with Occulus Rift programming through Unity. Not that you want to be in games development, but, as far as technology goes, VR is just cool.
Also, I'm a believer in Turing completeness (even if HR resume screeners aren't).
I'm a big believer in this being nonsense. I see no reason to think that someone who's a great Python web dev would be very good at writing Unlambda.
That is, it may well be true that learning one mainstream or even non-mainstream language will make it easy to pick up other mainstream languages later, but that has zero to do with their being Turing complete.
If you do that well, you can write/use the trivial stupid algorithm and substitute in a better one as you find you need to.
If only there were a well-known pithy statement that encapsulated this bit of knowledge!
(I kid.)
My metaphor crumbles before the superior size of your languages dick. Actually.
My understanding is those are functionally equivalent regardless of size.
Jesus, Taiwan's about to get nailed by a Category 4.
Then in a year you can spend six weeks playing with Numpy and call yourself a 'data scientist' and get a 40% raise.
Not that this isn't almost exactly what I did, but you need to know some linear algebra and statistics.
50/51: That stood out to me as well, but I thought it was a joke trying to say that HR screeners could be replaced with e.g. regexes.
I stand by the usage in 43 as a useful shorthand for the idea that since all the languages are capable of doing the same things, we can expect a lot of convergence among mainstream languages around common idioms and therefore that skills in one language are more or less transferable. It's true that said convergence isn't entailed, and maybe it's foolish to expect it, but we do in fact observe it. Lots of HR departments don't see it that way and filter by keyword so it can valuable to be able to list those keywords explicitly for job-hunting, but I wasn't trying to make some comparability joke about that.
It's also true that the convergence described doesn't hold for non-mainstream languages, but if you work for a pace that uses Unlambda in production and is big enough and growing fast enough to have an HR department to review resumes ... congratulations? Maybe you can pitch Ogged on learning that one special.
It is basically not possible to use Unlambda in production because (in addition to being a write-only language), the sole concession to the idea of the real world that it makes is the ability to do byte-by-byte I/O to and from stdout bzw. stdin. You cannot open a file, access the network, make a system call, make a library call, use a thread, or anything like that. (Also, there are no built-in numbers, strings, sequences, or any data structures at all other functions, so you have to use a Church encoding or other like encoding. Have fun implementing floating point stuff, I guess.) This is one of the reasons that the "Turing complete" shorthand is something of a hobbyhorse for me, actually: just being Turing complete gives you a language that is sublimely useless.
59: I agree on the principal that most mainstream languages are similar and skills are transferrable, but I don't think saying that because they share that one mathematical property is sufficient. It wasn't that long ago that arcade and console games were written entirely in assembly. I could learn to do that, but it'd be hard and time-consuming. For that matter, I think I'd have trouble working in a codebase of a programming language I knew where all the identifiers and comments were in a natural language I didn't know.
So I don't think Turing completeness is sufficient; I think it's more due to the social practice. Today's languages' syntaxes--and for that matter their semantics, too--derive from a small number of antecedents, and the people who make languages are comfortable with them so they perpetuate those. (Excepting esoteric languages, of course.) As another example, when people who have learned only C-like languages are first exposed to a really functional language like an ML or Haskell, they tend to have a hell of a time.
But, uh, your general argument about how to interact with HR departments applies so I'm mainly being a silly pedant.
60: so what we need is 1) a way to provide access to wrapped C libraries in Unlambda and 2) a compiler from a real language. That'd be a fun horrible project.
I believe somebody made a compiler whose target language was Mondrian. Its output was beautiful.
I have several Unlambda interpreters.
The Rust and Haskell ones will certainly not build as-is, though, since I didn't bother including any dependency information (and it's pre-1.0 Rust).
Is this the GOP debate thread? Smearcase and I are watching from the edge of the lower 48 American west.
59. A computer language and its adherents typically embody a way of thinking about problems, which is a separate thing from whether it is Turing complete. For (simple) example, OO languages are about "everything is an object." Functional languages are about mathematical formalisms, caricatured as "nothing has side effects." Etc. That's not even to get into the "culture" part of a language community, which is significantly about the ways the libraries and development tools shape the things you can do with it.
So, yeah, "it's a Turing machine" is true, but it's also a joke, a la "it's a simple matter of software." (For those not in the computer biz, the irony level in both statements is very high.)
Neat! I've written one too, in Haskell. Props to the Unlambda people (well, person, probably) for making it so easy to do.
I was thinking about trying Rust, for fun--what are your thoughts? Is it stable enough to give it a try?
Jeb just used up all his talking points in 20 seconds.
Today will be the day I learn how to work the "Remember personal info?" checkbox.
Also, "I have several Unlambda interpreters" would be a great debate counterpoint.
Rust is certainly the most cumbersome of any language I've used recently (Rust, Haskell, Clojure, Python), which is unsurprising given that it's the lowest-level, but I like it. I haven't done much with it—I implemented a functional red-black tree with cursor-based traversals, and I did some of the Matasano crypto challenges with it, and that's about it. I did enjoy using it and it seems quite well thought out.
61. Check out some languages uses in Formal Methods. ACL2, for example, looks like it's Lisp but takes a long time to open your mind to what "code" in it really means.
61. Check out some languages uses in Formal Methods. ACL2, for example, looks like it's Lisp but takes a long time to open your mind to what "code" in it really means.
It's so complex it makes you double post! ObExcuse: I pressed the button just once! I swear!
73: I guess that's enough of a recommendation for me. I've been too removed from languages at that level and I've been meaning to do some work in one. (In my opinion Java is intermediate between languages that don't do much to abstract memory and functional languages.)
And as long as we're talking about awful-but-Turing-complete languages: I once wrote a program to output the Fibonacci numbers in Whirl. It took me several days, and that's without cleaning up memory.
Why are you watching those clowns? Pure amusement, or is there some actual reason to?
I'm not watching, only because I have a choice between sitting in front of the TV or sitting in a place with decent Wi-Fi reception. Fucking around on the Internet won out over watching a bunch of horrible people say a bunch of horrible things.
It's pretty entertaining. And Smearcase watches the debates.
Do any if you follow @KStreetHipster on Twitter? Her series of tweets this morning about how the Dems should have as many debates as the Republicans was sublimely frustrating.
Urple was trying to do the debate on another thread.
My son can get to the very top in Doodle Jump. I think that means Google will hire him in ten or twelve years.
I realized that 1) I'm not working tomorrow and 2) I have whisky so I switched to the debate. Having a good horrid time.
Donald is a comedic genius. I mean a delusional narcissist. Anyway everyone is ridiculous and it's pretty fun.
Also the moderators are doing a much more stand up job than I expected
91: so is he sunni or Shi'ite?
Must be one of them no go zones I've heard so much about.
What's the principle whereby, if there's an old sitcom you only watched/remember one episode of, if you come by a randomly playing TV, and it is that show, it will also be that episode?
This is a variant of the plate o shrimp phenomenon.
95: I used to call it the "Does the United Kingdom have only one episode of 'Nash Bridges' that they play again and again?" principle.
Modified version of caine-hackman conjecture.
trumpy is the worst of the specialized python distributions
Paul: Haskell
Kasich: C++
Jeb: COBOL
Carson: MS Word
Cruz: Javascript
Christie: SML/NJ
Huckabee: I dunno, fuck that guy.
Christie: SML/NJ
Surely Christie would embrace "worse is better" and go with C?
Look, if you know a programming language that comes closer to saying "I wish I were a Springsteen song," let me know.
The, er, NJ bit? God, I fail at jokes.
Right, I got that, but C is also from NJ. And SML/NJ is just one particular implementation, right?
If we allow implementation names, though, a whole world opens up (especially among Schemes): Scheme itself, Racket, Gambit, Stalin, Chicken … And the ML compiler MLton.
Well, I was responding to a joke where Trump was analogized to a particular distribution, and I at no point specified the type of my joke objects, so I think you need to get your type inference engine checked.
But getting back to implementations, I think Christie showed that he's a particularly good implementation of the GNU/Giuliani specification.
Huh, half of those Scheme impl names are fighting game characters. (Stalin is basically Zangief, right?)
Paul: the pi-calculus.
Kasich: double-barrelled continuation-passing style
Jeb: System F
Carson: ordered linear type system
Cruz: unbounded nondeterminism
Christie: the sound of one hand clapping
Huckabee: 80386 assembler
This seems to be a minority opinion, but I don't think that language matters much. I mean that both in the sense that I've found the commonly given advice that you should learn new languages in order to learn new paradigms to be unhelpful and in the sense that I've literally never been hired for a job where I've been proficient in the primary language used in the job.
FWIW, the languages that I've written > 1000 LOC in professionally are C, C++, various flavors of assembly, various flavors of microcode, Verilog/SV, F#, Go, Ruby, and X. X is redacted because it's a theorem proving language/system that's used by few enough places in industry that between that and the list of other things I've used, some of the X community would immediately know who I am.
Anyway, despite having used a variety of languages I've never found that learning a new language paradigm has really helped me much, and I've gotten much more leverage from my non-language skills.
Having said all that, my impression is that for a large fraction of entry-level jobs (web dev, internal enterprise apps, maybe some others), knowing a specific set of tools is required to get your foot in the door. For enterprise stuff, it's often either Java or C#. For web dev, it's what the trendy bootcamps are teaching (often Ruby+Rails, though js and Python+??? are also options). If I had to guess, I would say that the easiest path would be to get one of those somewhat specialized entry-level positions, and then branch out afterwards.
73: I find Rust to be much more cumbersome than C++. I'm not sure if that's because it's actually more cumbersome or if it's because I know C++ better.
Python+???
Python plus Django or Pylons/Pyramid seem to be the most common. I use much lighter web frameworks as what I'm doing is largely backend stuff, or building shims that translate between different APIs, so there isn't a UI component.
Ogged,
Ok, with more information, other suggestions. You said that you want a "course of study wiith a defined end-date". This, to me, says "take a course". And today, there are many excellent courses available. Can I suggest:
(1) the MIT course -- I believe it's the modern equiv of 6.001, which was (back in the day) the first CS course MIT frosh took. And was an excellent, excellent course (I was never an MIT student, sadly, but many other schools used the lovely SICP book; again, this about the Scheme version of the course, I'm sure the current version is quite different, but still, MIT, frosh course, can't be bad).
(2) take a course in basic data structures and algorithms. Not the fancy algorithms stuff. Just what sophomores take (sorting, balanced trees, stuff like that).
(3) And take a course in Java
I know this is a lot. Maybe too much. If it's too much, then take the Java course -after- the first two. Here's the rationale: you want to know a language that can get you employed. Python can be that language. But you also want to get started on actually knowing programming at a level where you don't have to constantly keep running after the latest trend. and that's where starting (-starting-) to understand datastructures and algorithms will help.
So: why not do #2, #3? What's so important about #1? Well, again, I think it's important to not just learn the "incantations" of programming, but also to learn a -bit- of the deep knowledge. And I trust that that MIT course will impart of the deep knowledge.
Again, I think that if you're going to invest time and energy in learning this skill, it's worth spending -part- of that investment in a manner that doesn't degrade over time. Hence, #2, and part of #1. And learning Python isn't bad as a skill for getting a job.
Also, as an aside, while learning these things, you might want to learn to become a decent sysadmin -- on you own machines. This isn't difficult either, and you can get there by baby-steps. Start with Ubuntu, and then move on to Debian (b/c Ubuntu sugar-coats a lot of Debian). Learning how to be a decent (not excellent -- just decent) sysadmin is critical to being an excellent programmer.
OK: so that's even -more- work. So again, I think the best order would be what I listed above: #1, then #2, then #3.
I cannot sing the praises of the old 6.001 course enough. And I have to believe that Sussman (who designed the old and new courses) wasn't completely foolish in how he designed the new course, if only because he really is an excellent educator. That's really the biggest reason for advocating the course: b/c it comes from a pedigree of proven educational value.
Reading 115, I think he has a point. If your goal is to get a job -first-, one of these bootcamps might really be the trick. He's 100% right, that for many entry-level jobs, knowing the various frameworks and toolkits du jour, is more important than really knowing CS well. He's also right that one -can- learning more deeply after getting that entry-level job.
It's possible that you'll try one of those bootcamps and feel completely at sea. In which case, backing off and actually taking a course to learn programming in a more traditional path can help with that. But a -lot- of folks got into programming in the manner that 115 suggests -- start with Perl CGI scripts, and end hacking C++ into the depths of Windows kernel drivers, N years later.
It does require a bit of bravery at first, b/c you feel like you're the only one faking it. But actually, many, many people fake it, so why not you?
The approach I've tried to take, who knows if it's right, is splitting my time roughly between learning
65% - Current technologies and tools
35% - CS fundamentals
For the CS fundamentals, I took some old-fashioned university classes. I was lucky in that an over-qualified young MIT PhD happened to be teaching community college classes locally for a few years (she's since gone back to building robots). I am trying to go through MIT's XSeries in Foundations of Computer Science (scroll down), but they're slow in publishing the courses.
I invested in the bootcamp mostly for the career services, to overcome impostor syndrome, and to get the experience of pairing and working in teams that I couldn't get through self-teaching (or really from standard university pedagogy).
108. Even better, Scheme was an outgrowth of Planner and Conniver (or maybe Conniver was after Scheme?). Anyway, language names at MIT were often jokes: for example, another Lisp outgrowth was Muddle.
115. Python+Javascript, maybe? What a lot of web dev employers are looking for is someone who knows their current favorite Javascript framework. Since there are about 8000 of those, you can't predict what they want (unless they say in the posting). I think Backbone is still pretty strong, as is AngularJS, but there are plenty more. Also learn some of the libraries; jQuery is the most widely used.
118. On faking it. When I interviewed for my first non-university job, I had to have C, so I read K&R over the weekend, was honest that "I didn't have too much C experience," and got hired.
Wait, people use theorem proving systems in industry? I thought they were almost purely academic ideas? (A friend did a fair bit of work on them, but all from the academic side.)
121. Yes, they do. Yes, but things actually move from the academy to industry now and then. Shocking, eh?
121: One common application is verifying that computer chips work correctly. Here's a discussion of how they use it at Intel.
Thanks again. This is what I'm thinking after having read the thread. I can't do a bootcamp because of childcare constraints. There's a CS certificate program at a local school (to which I've already been accepted) that's all C#, but they offer a compressed Intro to OOP/Data Structures course in the upcoming semester, and I could take intermediate OOP and a more broad "Programming Languages" course next semester (as well as a couple other courses to round out the requirements). The courses are taught by regular CS faculty, and I figure that should give me the "theoretical" base I need, and though my first job will likely be a .NET job [shudder], I should be able to branch out fairly easily (for those not keeping up at home: I've made one app in Objective C, and have been working through a Java book on my own).
Does this sound reasonable?
If you aren't going to learn SAS, I guess. But it's worth pointing out knowing SAS is my most marketable skill and I've earned a living on it despite having nothing much but that and white privilege going for me.
123 - yeah - I mean it's the kind of work that if it is relevant to anyone. it's relevant to everyone...
There's a similar CS program at a school further south, too, you know.
(he said as it were - (ahem) bluntly (ahem).
127: looks to be a much longer program. plus it's faaar.
C# is definitely MS-only, despite the existence of Mono.
but, as others have already said, Java and C# are pretty similar conceptually. if you already know one, working with the other will be like cooking in someone else's kitchen - it will take a little work to find out where all the utensils and ingredients are kept, and you'll miss your favorite special spatula. but you can get the job done after a little puttering around.
Speaking of this kind of stuff, if you ever convinced anyone in my company to buy crappy, hard-to-update, Java-based commercial software I am trying to kill you with my mind right now. Also if you ever thought, "Hey! This Silverlight stuff is great!" and subsequently acted on it.
Hard to update commercial Java-based software works best when it's a client for an Oracle database and the preferred way to export data is via another, separate reporting application made by a different company that's designed for formatting documents.
Oracle databases don't work well with commercial software the updates easily?
My operating delusional daydream is that as I ease myself into retirement I will embark on some manner of Python and Javascript programming. I almost certainly lack the ambition and drive, but I have a number of projects that I dream of whilst woolgathering. For instance, instrumentation and stats--and ultimately an AI--for DIVE, the Threes/2048 variant that I am still addicted to.
Colorless green oracle databases update furiously.
138.
What is your high score, and have you started running your own variants where you fiddle with addRandomTile() ?
6843 here
I have a computer question. Is it possible to accidentally set a nickname for a google account? If not, is it possible for a google account to be hacked with 2-step notification enabled and having never received a text?
Anyway, I think it must have happened today as I'm pretty sure someone would have mentioned it if I had been showing up on meeting notices as "GlobalMediaConspiracy."
24821 (see screenshots here).
Not done any fiddling, nor do I think I know how.
That result is a long tail, for a long time I was stuck just beneath 10000, then broke through with that score and shortly there after a 23K. But otherwise probably a dozen in the 10K - 16K range. My "strategy" is to basically live with all of the factors up to and including 13; and try to eliminate everything else that happens except concentrating on a single "build" which I try to get it up to a big (three digits) prime as quickly as possible and milk that (on the screenshot you can see I got into a nice run of x99s/x01/x03--199, 299, 399, 599, 799, 601, 1201) . However, I am not convinced how much that adds; and suspect my results are only marginally better than random (part of what I want to test by building one myself), and all you are seeing is the result of me playing it a lot more times (it has become my fidget--I play it on the phone in all semi-downtime).
I had thought that there was a randomization problem, since most recently added factor shows up pretty often. Not convinced it's working right, but I basically dislike js, so not doing anything.
The whole thing's in javascript, calls to Math.random() are apparently seeded with time, or whatever the browser's js interpreter provides for that.
https://github.com/alexfink/alexfink.github.io/blob/master/dive/js/game_manager.js
suspect my results are only marginally better than random
https://en.wikipedia.org/wiki/Gumbel_distribution
141 was serious, if anybody knows anything useful.
||
Help the braindead idiot, gratuitous insults ok
Time Warner been upgrading lately
I am downloading, with my own throttle, at 750 kBs
My wireless connection shows bandwidth of 72 Mbs
I can't seem to trust or make sense of the conversion charts.
|>
Moby: https://support.google.com/mail/answer/45938?hl=en
125: Don't you also have some kind of stats background? My (possibly incorrect) impression was that you'd need SAS + stats skills, and that only knowing SAS (or R or whatever) without a stats background wouldn't be helpful.
124: That sounds generally reasonable. This is probably well known to more regular commenters, but what does the local tech job market in your area look like? I don't drop in here often enough to have any idea where you live, and the demand for specific skills can vary wildly by region.
Also, have you tried online courses? I don't think they'll help you get a job, but they can round out your background if that's something you're looking for. Just for example, I've heard that both sets of algorithms courses on coursera are quite good (there's one sequence from a Princeton prof and another from a Stanford prof), and that the Grossman PL class and this Caltech ML class are excellent.
I'm in the hog butcher to the world metropolitan area.
Thanks. I'll look when I'm soberer.
Yes, I do know stats. But that's just common sense.
149: Unfortunately, every programmer I know in the area either works in finance, at Google, or remote for a startup, and I have no idea what the job market there looks like. I can ask around if you think it would be helpful, though IIRC there's someone here who works at ci/vis. That person probably already knows what the local tech job market looks like.
I watched the vuds for the Stanford algorithms one. It was interesting. If I'd had the time and skill to program along with it, it might have been very good, indeed.
143: I had thought that there was a randomization problem, since most recently added factor shows up pretty often
I had previously downloaded and looked at the javascript. And in fact what prompted me to do so was a conviction that 13 (back before my current strategy when I regarded it as evil) was coming up too frequently ... So I suspect there is some manner of cognitive bias in perceived frequency of unwanted outcomes that we both fell prey too.
If you look for a thread on DIVE starting March 14, 2014 on the guy who wrote it's Google+ (there must be a way to link a single post, but I'm not seeing how) you can see some discussion on it (and some earlier threads). And when I checked today, I saw that he and someone else seem to still be playing it, and in early July he mentioned his high score was 13772.
Ah, here's the link to the post/thread mentioned in 153. 4 recent comments. It's clear he spent a lot of time on the color/patterns and I think they work pretty well.
148.1: Just the data management part of SAS should be enough, for those bereft of common sense.
153.3: The date of the post is the link to that single post.
I think this is a neat bit of Rust type manipulation, albeit it not very sophisticated compared to this madness.