Learning to Code: The Roadmap I Wish I Had Been Given

If I could add a stage though, it would probably be something like "Cluster 5: Why does my website suck?" Sure it's there, it's up on the WWW, but a website made from someone at Cluster 4 isn't something that is going to make you money.

Cluster 5 is when all the CS comes in. You have to learn about algorithms and data structures and regular expressions and what is actually going on behind the scenes. Yes, you can make a website that does the basic simple "of course, here's an exhaustive search" solution to the given problem. But can you make it fast? Or beautiful? Or, heaven forbid, intuitive? Can you fix it after you try and fail to make it the above three? That's the stage that never ends, the one that doesn't take a week of hunkering down with a textbook to get through.

Cluster 5 is the stage that CS education tries to start everybody at without any prior experience. It works for the imaginative people who can make the leaps about why this matters, but for most it just leaves everyone behind. With web programming, you actually have motivation to learn these weird things about patterns of electrons.

Because, honestly, who wants to make websites that suck?

As for names, at SFU (my alma mater) it's called "Formal Languages and Automata". Well, actually, there are actually about half a dozen courses covering parts of this material, but that's the most-perfect match for what you described.

     Learning to break apart a regex and rebuild
     it as a state machine is where the real magic happens.

That is very true for CRUD apps that make up the majority of web apps. And if you need to do something more complex with a web app, like analyzing data, the CS knowledge required is no different than if it was an offline app, i.e., a CS degree would be very helpful.

However, I think that for the majority of programming jobs out there, a full scale CS degree is a bit overkill.

If you're hiring someone to produce CAD drawings, you don't need a full fledged engineer or architect; you can find someone with a certificate from a tech school.

I wonder if eventually programming will eventually go that route. 4 year degrees for Software Engineers, and tech school certs for Software Technicians.

Another common example would be a course in number theory. A common topic in number theory is public-key encryption, much more in-depth than is sometimes covered in other courses. In addition, the techniques that you learn in such a course can easily be applied to any "number-crunching" you would need to do. Here is an example:

Problem 1 from Project Euler (http://projecteuler.net/problem=1): Find the sum of all the multiples of 3 or 5 below 1000. A brute-force program for this is simple, but it is much faster if we first notice some patterns: (3 + 6 + 9 + ... + 999) = 3 * (1 + 2 + ... + 333). But then (1 + 2 + ... + 333) = 334333/2 (this would be easily recognized by someone who has taken a number theory class), and so the sum of all of the multiples of 3 below 1000 is 3 333 * 334 / 2. Similarly, for 5, there are 199 multiples, so the sum of all of these is 5 * 199 * 200 / 2. But then we must subtract the multiples of 15, since we double counted them: there are floor(1000/15) = 66, with sum 15 * 66 * 67 / 2. So the answer is (3 * 333 * 334 + 5 * 199 * 200 + 15 * 66 * 67)/2 = 299498.

That was a particularly trivial example - number theory will give you a lot more than some parlor tricks like the above. But this example took a loop through n numbers and reduced it to a single line of simple arithmetic. And for someone who understands number theory, this algorithm is as quick to think of as it is for a typical CS student to think of the brute-force algorithm.

I would also venture that a course or two in "Abstract Algebra" would be both beneficial and accessible to CS majors. This is less immediately practical than linear algebra in the above example, but closer in its style of thought to programming than, say, calculus, topology, or differential equations. The abstraction (which many math majors complain about) would be fairly straight forward for a CS major, I would think.

Don't install Ruby libraries through apt-get, and also don't install them through RubyGems globally (i.e. never use "sudo" for installing gems).

Instead install gems for the local user only, with the help of RVM. Then to upgrade, if everything breaks, you can just delete your RVM directory and start from scratch again.

Of course, this brings with it a whole other can of worms, because then it's your responsibility for upgrading your libraries with the latest security-related fixes (which otherwise would have been updated with a simple "aptitude safe-upgrade", which can be a cronjob or something). But then you've got the same responsibility on Heroku.

See http://beginrescueend.com/

jQuery's a great choice for beginners learning to program for the first time. Two related frameworks, underscore and d3, both have jQuery-inspired APIs and use higher-order functions extensively. These frameworks teach lessons in functional programming that transfer easily to Python, Lisp, Haskell, and even mathematics.

Coding/programming encompasses many programming languages, frameworks, platforms, etc.

Web applications use a subset of these, by using mostly markup languages (html/css) and some web languages (javascript/php). They are deployed to servers with http software, and are built for handling web traffic.

The web app itself is typically going to be running on a framework, written in another language - for example, a site might use Rails, which is written in Ruby.

This has to be served by a webserver - Apache is a popular choice - I think much of that is written in Java.

The webserver runs on a machine, which might be running, say, Linux, as its operating system. The Linux kernel is written in C.

Now the page has to be transmitted across the network - hopping across the globe from one router to the next. The hardware that does all this switching is in microchips, which are often designed in a language called Verilog. (This is my own little niche area)

The browser that you're using to view the page might be running on a laptop running Windows, which is largely written in C++.

There are countless other fields of "coding", too.

So to summarize, writing web apps is a subset of a vast and complex field of study - and to follow this roadmap (which is probably a good way to learn what the author wanted to learn) will not teach you anything about most of the other things one might want to code.

For what it's worth, in setting out on your journey to learn about programming (in the broader sense), I think you could do worse than to start with C and Scheme.