What we read in February: Angular sources, results of the year in numbers, frame recognition and other useful links

We have more than a thousand employees in different departments and with different interests. Sometimes they share useful, interesting or just funny links with each other in the corporate slack. I spent a month monitoring our main channels, and gathered the most interesting of them - I got an experimental team hodgepodge of recommendations from development, marketing, analytics, media services and management. I hope that readers of Habra will find something useful here for the weekend and for themselves.

Colleagues from the development department were impressed by the transition to the new Angular , so this channel appeared in the channel # dev-learn:

Clipcode source overview Angular 5.2

The Irishman Eamon O'Tuathail did a good job and wrote a 200-page review guide on Angular packages and their source code. In it, you can get information about how it works. Of course, we did not master the entire work, but bookmarked it - a useful reference piece just in case. Well, the text is a bit more interesting and simpler - a preliminary review of the upcoming beta in ng6 Ivy renderer . What it is, what it does, how it does what it will give (a small part is described).

Alexander Makarov with his recommendations was laconic:

Lasagna code: too many layers?

Another reminder that following any advice you can’t bring to extremism, and in general you need to think. On the example of layers in application architecture.


Paradoxically, I noticed that for understanding why, when and how to break the rules, it helps to follow these rules very strictly for some time. When you feel that going against the rules is normal, document (and sometimes review) your decision.

Practical introduction to container terminology

One of the best Linux container reviews and how to work with them.


Even worse, many of these terms are used interchangeably when used, often causing confusion for beginners. <...> Understanding the terminology covered in this technical dictionary will help you better understand the underlying technologies of containers. You will be able to speak the same language with the team, and it is also better to design a container environment for the tasks you have.

Our marketing continued to count the past year and threw a couple of interesting topics:

2017 in numbers

A suitable selection of interesting facts and figures of the past year. The article helps to understand the main trends in online advertising, e-commerce, voice technology, Chinese business, media and other industries.


BuzzFeed claims that its morning show 'AM to DM' is watched by up to one million viewers daily. In comparison, the daily audience of Fox News’s largest cable network is 1.465 million people ( TechCrunch ).

One in six Americans, or about 39 million people, have a “smart audio speaker.” This is 128% more than last year ( TechCrunch ).

On average, only 0.06% of page visitors click on banners, and about half of them do it by accident ( Google RMG ).

The battle for consumers goes online to real life

Airbnb buys real estate ... Uber buys cars ... Amazon builds stores ... Netflix shoots TV shows ... An interesting article about the fact that all the companies that worked on the model of offering an online interface for offline goods and services have a huge estimate, but are unprofitable. To support this assessment and start earning money, they begin to fulfill the functions of those they aggregate themselves. And with their data on the preferences of millions of users, they make it more effective than old offline players.


It turned out that Uber wants less responsibility to employers than its own fleet of cars. In fact, the payment of 70% of the revenue is so close to financial insolvency that many consider the Uber business model organically unsuitable - unless, of course, the company gets rid of the drivers.

Dark times have come for many workers. Uber drivers (like warehouse workers) literally create training programs and test a business model in which they can find a replacement. The horizontal movements of Uber - Uber Rush, Uber Eats and expansion into Asia and Africa - did not bear fruit, so we can expect the company to develop its cars, make autopilot a reality and find other ways to become profitable. She will have to move in a vertical direction.

With the submission of our head of media services, a suitable article from the Netflix blog:

AVA: The Art and Science of Image Recognition at Netflix

All successful companies (including Skyeng) strive to put processes that have traditionally been done manually into the flow, to automate the work as much as possible to make it more predictable (and economical). In this post, Netflix talks about its algorithm for finding good frames (aka screenshots) to attract the attention of the public in the directory. Technical details promise next time.


AVA - a set of tools and algorithms designed to extract high-quality images from video. One season of the series (about 10 episodes) on average contains almost 9 million frames. Asking editors to go through everything to find the one that catches the attention of the audience best - boring and inefficient. We decided to develop a tool that can quickly find frames containing the most suitable scenes for representing the series in the Netflix catalog.

From him is a link to a controversial article worthy of discussion:

MVP is dead, long live MAP!

The author considers MVP “dead” and offers an alternative approach - not a minimally useful product, but a minimal, charmingly incredible (that is, perfect) product. He may have a higher entry threshold, but also a better chance of survival. It sounds utopian, but like a fresh look, which conflicts with the whole order of boring processes, it reads well.


It’s hard to do something good, beautiful and cheap. When developing MVP, you have to prioritize. Cost should be low, in the end it’s just a “test”, but ... Are potential customers ready to lose quality (or at least the appearance of quality) just because it is a test?

Managing partner Alexander Laryanovsky recommends a book read in February:

The Brain Healing Himself (Norman Dodge)

“The ability to control your brain is the ability to manage mood, performance, concentration, relaxation. Of course, we cannot control the brain 100%, but we can learn to control a lot. So, to be at the right moment what you want. I recommend the book to anyone who thinks about their own effectiveness and productivity. ”

And the head of our analytical department Gleb Sologub recommends this one:

The Hungry City: How Food Defines Our Life (Carolyn Steele)
(English site)
“The book is very cool, it makes sense to read it to everyone, to imagine what kind of world we live in, where our place in history is, how our life works, how much going on behind the scenes of serving our daily needs. Many successful leaders, when asked what they read, call books of this kind that significantly push the boundaries of knowledge. Another of these is Sapiens. A Brief History of Humanity and Mining: A World History of the Struggle for Oil, Money, and Power . ”

And finally, taking advantage of the official position - a link from me personally.

Self-taught Programmer Stories

While preparing an article about self-study programming, I came across this archive thread, which stuck for a long time. Unlike refined “success stories” from advanced bloggers and coding schools, the honest experience of ordinary Reddit users is presented here, without embellishment. And the English way is not always perfect, but you can get an accurate idea of ​​how much time it takes to learn, how much effort you will have to put in and what kind of work you can count on as a result (spoiler: it's not so bad, but you have to be ready to moderate your appetite).


I have been interested in programming since childhood. By high school, I already made websites and all sorts of other things. But they told me that in order to become a programmer, I have to love mathematics, so I went to college and got a degree in writing. I really started to learn C #, getting a full-time job as a technical writer and realizing how terrible this work is.

Let me remind you that we have a translator extension for Chrome - it’s easier and more convenient to read English-language articles with it.