Nice intro to Linux commands. I don't know why, but for creating new files I prefer redirecting with echo > file.ext . With cat problem is that textual information will be copied into destination, but maybe file is binary ? So either your proposed touch or echo with zero bytes outputed. Then you open that binary file in required edit, and here you go.

Nice intro for beginners. Should be good to mention that there is not several, but at least 700 programming languages according to wikipedia. If all dead languages to be included, which nobody has remember it now,- probably we could get 10x bigger factor of programming languages. Also in nowadays boundary between compiled and interpreted languages is not so clear. There can be niche where both processes takes place in running code. Even in C#, Java there is a two-step compilation, at first source code is compiled into intermediate virtual machine byte code, and then this byte code is interpreted on-the-fly, and WHEN needed JIT compiler compiles part of it into machine code. Python on the other hand is interpreted, but there exist tools which can build standalone machine executable from Python source code. So what I wanted to say from this mess that boundary in nowadays between compiled/interpreted is actually blurred. But maybe my notes would be too much pain for beginner minds, so maybe these were left intentionally for curious self-learner :-)

Brilliant article. As I came from very old pre-agile ages, when the only curse was "waterfall" and tons of semi-blank paper of documentation - I can see some evolution of management processes. Then came agile, which was more democratic in respect of developer needs and technical expertise and free-will. Now we see, that still agile has some drawbacks and it is not fully goodwill-compatible with developer. It still pushes/forces some weird things to developer against his/her will. That's why we have post-agile movement now started. It's correctly noticed that self-organisation is main keyword here. Developers are not stupid, they knows what must be done roughly to the client, so best thing would be to let them all self-organize, because only they knows best team expertise / wishes / likes-dislikes / tendencies / capacities / etc.., etc. Managers pushing they own perception of work process automatically does big hazard to work quality and team efficiency. Everybody knows that best things are done by people who accomplishes something they LIKE and in a GOODWILL and in a dead-lines they themselves see reasonable. We are not robots, so we can't work as a robot. And every (forcible) work process makes workers robots to some degree. If there must be some managers - they must be just as " advisory force " and not like "control force". In the sense that we must stop lying to ourselves that we can "control" something in principle. Actually - we CAN'T. We just can assemble brilliant people in a team, setting transparent goal for them and let the team do the rest in hopes that team will have good self-organization and that it will reach required business goal. Nothing more, nothing less. I would call this post-agile management process as " goodwillchitecture ", instead of pointing something related to anarchy. Because in general anarchy is perceived as bad, negative thing with no rules or order at all. But self-organization DOES have rules. The most important one is - goodwill, tendency of team to make a great positive thing towards a common goal. That I would not call anarchy.

Nice article, I especially liked part about HTTP/2 multiplexing, indeed - every web server should make this requirement a MUST HAVE . However I can't agree on your note that : Using the Switch case over if/else enables developers to shorten execution time I can't believe in this - at a lower level both code structures should be compiled into some sort of goto conditional jumps. Let's check. I've made a JS benchmark. This is switch block: var rnd = Math .floor( Math .random() * 5 ); switch (rnd) { case 1 : rnd += 1 ; break ; case 2 : rnd += 2 ; break ; default : rnd += -1 ; } and this is if/else block : var rnd = Math .floor( Math .random() * 5 ); if (rnd == 1 ) { rnd += 1 ; } else if (rnd == 2 ) { rnd += 2 ; } else { rnd += -1 ; } After pasting these blocks into jsbench.me performance comparison suite,- I've got results that switch block is SLOWER by [0% - 2%] on average. Given many unknown factors - how these tests are executed, JS implementation differences between browsers, execution speed measurement errors, biased outcome (maybe after 1000 or more runs performance difference would be 0%) and etc. - it would be rational to neglect these small execution time differences and for practical purposes keep that both should run at the same rate.

Nice article. I would like to add that not every thing can be unit-tested with a test automation framework. At least, not practically- if your single test would run for a several days . Consider this, that you are building software which categorizes movies by genre. You can't write a unit test for just a several movies and if they are correctly categorized - hope that your algorithm predicts genre well. It's not working like so. Typically you would want to calculate false positives and false negatives for a genre detection of let's say 2'000 movies. If you will get these false positives / false negatives ratios as a small numbers from this movie set - only then you can conclude that your algo works as expected and change X in the program hasn't broke the classification quality. The problem is that classifying such huge amount of movies requires a long time. So it's reasonable to put such classification task in some sort of CRON scheduler which would be run regularly until it finishes classifying a movie test set. Only then you can calculate FP/FN ratios and say if your test has passed or not. What I'm trying to expose here - that long-running tests may not be suitable for a automation frameworks .

As I've understood from the survey, main question was - will we benefit from a custom platform which targets UI things (animations / components) selling ? Don't think so. It's of course cool, BUT : There are plenty of sites already which exposes designers cool stuff as a portfolio, for example codepen.io , very mature and huge number of submissions and free Creation process is not the same as baking buns in a factory. It comes spontaneously and also a very hard work required. So unique and usable / sell-able things will not come at a rate from which you could get a constant stream of money for a living Even then if you come to a thing which you will feel will be highly profitable - it's almost likely that you will sell that ONLY thing alone in it's own purchased unique domain with it's own unique design.

PHP is not the best language according to professional developers Well, you may consider me as a noob in web development, but PHP is object-oriented, functional, dynamic, fast language (at least faster than Python). What else do you need for a RAPID web development ? This is the PHP philosophy.

Depends on what language,platform(s) you are using. For a web development it's enough of print_r() or var_dump() statements. In my experience of 2.5 years of PHP web development - I have NEVER used debugging facilities- just these statements above. IMHO, if you feel the need of debugging (in a web development) this means that something with your code structure is not right and you need a complete code re-write. Then sit down, and re-think deeply how you can help yourself in the future by changing code architecture now . About the question part - How and where to put breakpoints or print_r() statements in the code. Well, actually nobody can answer this question - just YOU. Because you are the master of your code and just you know it well (except maybe your colleagues or a project manager). Analyzing code paths is a developer duty and we must perform it well. So get used to it, otherwise it may mean that developer job is not a dream work for you.

Nice article. I will admit, that I was a bit skeptical that Java uses Tracing garbage collection instead of traditional reference counting. So I made a quick test to see if java will not get Out-Of-Memory exception when bunch of cyclic objects will be constantly generated in a loop : public class gctest { private gctest self; public gctest ( ) { this .self = this ; } public static void main ( String []args ) { while ( true ) { new gctest(); } } } Indeed Java program has not crashed and GC perfectly cleaned circular references. I only need to mention that traditional Reference counting GC model has also advanced and now is able to clean circular object references too, because there exists special algorithms in RC model which searches for a cycles in references. The same GC test in PHP (which uses reference counting GC model) : class GCtest { private $me; public function __construct () { $this->me = $this; } } while ( true ) { new GCtest(); } Script also has not crashed and traditional RC GC model cleaned objects with circular references. The only difference between Java and PHP in this case which as was able to notice - is that Java program has consumed ~ 100% of memory, but PHP version has not altered memory usage. Probably this means that Java GC collecting cycles starts at far lower rate than in PHP. Thus creation of objects fills heap at a rate at which Java GC collector is not able to catch-up.

Amazing post. I agree by 100%. Typical job posting for a web developer position looks like " It's mandatory for you to know 100 back-end frameworks and it would be good if you would know 50 front-end frameworks too ". I suspect the biggest reason of this phenomenon is that companies tries to cut costs and/or they simply don't finds suitable persons with skills required, so throws every possible duty to the hands of one worker. And the bad outcomes of such approach appears very quickly. For example - I have asked my manager why my company don't starts some project X. The reply from the manager was that X project was started in the past and it has failed. I have asked further - What solutions have you tried in a failed project X ? The reply was,- "the ONLY solution Y". After this answer I was amazed - How can one declare a project X as "failed" if there was only one solution Y tried ? It's not that hard to imagine that in IT field almost every problem has at least several solutions. So after trying N solutions you can only state that "exact N solutions don't helps to solve problem X". But maybe N+1 solution will solve that problem finally. BUT it takes only a professional in some field to actually be able to grasp and generate this N+1 solution. Jack of all trades will not be able to achieve this. As long as companies will not be able to understand this - we will swim in a deep sh*t sea as we always were.