A few months ago I bought a new Macbook. I’m really fond of the development environment I set up on it. The more I work with it, the more I love it. However, while building the environment I couldn’t find any decent step-by-step manual. So, it took me for a while to get what I wanted. Here’s my experience. Front-end environment For the beginning we need a package manager.
Automated testing is an essential part of application life-cycle (see Continuous Integration). In web-development we usually test: Back-end (*Unit-tests), Front-end (HTML/CSS validators, JSLint/JSHint, CSSLint, YSlow etc) and UI (functional testing). Unit-testing (Back-end, Front-end) is about checking if all the application components adhere the technical design requirements. Every unit-test simulates the application environment on which runs. Functional tests are to determine if the application as a whole ecosystem has no flaws. They run on a copy of real application environment and show if everything is ok from user prospective.
You know, when coupling is not loose, components depend too much on each other. It makes your entire architecture fragile and immobile. You can check how loose the coupling is by making a unit test for a component. If you have no problem substituting dependencies by e.g. mock objects then everything is ok. Let take a model class. It depends on DB connection, here \Lib\Db\Adapter\Interfaceinstance. We cannot just create DB adapter instance within model constructor, because it depends on configuration data which doesn’t belong to the model.
In my previous article among other patters I was telling about Flyweight. That is about the objects designed so to minimize memory use by sharing as much data as possible with other similar objects. You can find plenty of Flyweight implementation examples in Internet, though it will be mostly variations of Multiton, a collection (map) keeping only instance per every identical object. And I decided to follow that unwritten rule as it seems to be the simplest way to show the idea.
After having your project fully tested, deployed and running, it seems the application architecture is pretty good enough. All the requirements met and everybody is happy. But then as it happens, the requirements change and you, all of sudden, find yourself in the time of troubles. It comes out that some modules easier to hack than to modify. Change of other ones brings endless changes in a cascade of dependent modules.
Few years ago only lazy didn’t say about bringing desktop application experience to the web ones. However in reality, it just meant that user actions didn’t always required page reload, but could change page UI dynamically. As for other application events, as a rule they were not handled dynamically. Well, now you can find more and more web applications acting really like desktop ones. For example, Facebook and G+ have widgets which update automatically.
When testing a web-site, nevermind who you are developer or QA-engineer, it happens to you pretty often to fill-in form fields again and again. Boring, stupid work, but how to make sure the form does still work as intended? Some fields added, CAPTCHA was attached, whatever else done –you have to run the test again. Besides, it will be repeated on different browsers. Browser form auto-completion feature helps a bit, but that is not the same as when you have various sets of test-data always ready to apply on a form, isn’t it?