Hi guys, I work as a programmer for more than 15 years and was using many different languages, paradigms, frameworks and other shit. And I want to share with you my rules of writing good code.
- Optimization VS Readability.
Always write code that simple to read and which will be understandable for developers. Because time and resources that will be spent on hard readable code will be much higher than what you get from optimization.
If you need to make optimization, then make it like independent module with DI, with 100% test coverage and which will not be touched for at least one year.
2. Architecture first.
I saw many people who were saying “We need doing things fast we have no time on making architecture”. And about 99% of them got big problems because of such thinking.
Writing code without thinking of its architecture is useless in the same way as dreaming about your desires without a plan of achieving them. Before writing the first line of the code, you should understand what it will be doing, how, what it will use, how modules, services will work with each other, what structure will it have, how it will be tested and debugged, and how it will be updated.
3. Test coverage.
Tests are a good thing but they are not always affordable and make sense for the project.
When you need tests:
- When you are writing modules, micro-services which will be not touched for at least one month.
- When you are writing open-source code.
- When you are writing code that touches financial channels.
- When you have resources for updating tests at the same time as code was updated.
When you don’t need a test:
- When you are a startup.
- When you have a small team and code changing is fast.
- When you write scripts that can be simply tested manually by their output.
- Remember that code with badly written tests can be more harmful then code without tests.
4. Keep It Simple, Stupid.
Don’t write complex code. More it simpler then fewer bugs it may have and less time needed to debug them. Code should do only what it needs without tons of abstraction and other OOP shit (especially it concerns java developers) + 20% of things that may be needed in future to update it in a simple way.
To get in-depth knowledge on Python coding, you can enroll for live demo Python Online course
Comments showing bad code. Good code should be understandable without a line of comments. But what to do to save time for new developers? — Write simple inline documentation describing what and how method work. This will save much time for understanding and even more — it will give people more chances to come up with the better implementation of this method. And also it will be a good start for global code documentation.
6. Hard coupled VS Less Coupled.
Always try to use micro-service architecture. Monolithic software can run faster than micro-service software, but only in the context of one server.
Micro-services give you the possibility to distribute your soft efficiently not only on many servers but sometimes even on one machine(i mean process distribution).
7. Code reviews.
Code review can be as good as it can be bad.
You can organize code review only if you have a developer who understand 95% of the code and who can monitor all updates without wasting to much time. In another situation, it will be just time consuming and everyone will hate this.
On this part got too many questions so describe this more deeply. Many people think that code review it’s a good way of teaching new guys, or teammates who work on a different part of code. But the main target of code review it’s maintaining code quality, and not teaching. Let’s imagine that your team making code for controlling a cooling system for nuclear reactor, or space rocket engine. And you made huge mistake in very hard logic, and then you are giving this for code review to the new guy. How do you think what would be the risk of an accident? — On my practice more than 70%.
A good team is where each person has own role and responsibility for the exact piece of work. If someone wants to understand another piece of code then he goes to a person responsible for it and asks her. Impossible to know everything and better excellent understand a small piece of code than all but on 30%.
8. Refactoring does not work.
During my career, I heard many times “Don’t worry we will refactor it in future”. And in the future, this results in big technology debt or in deleting all the code and writing from scratch. So don’t get a debt unless you have money to develop your software a few times from scratch.
9. Don’t write code when you are tired or in a bad mood.
When developers tired they are making x2–5 more bugs and mistakes than when they are full of energy. So working more is very bad practice. That’s why more and more countries thinking about 6 hours workday, and some of them already have it. Mental work is not the same as working with your biceps.
10. Don’t write all at once — make developing iterative.
Before writing code analyze and predict, what your customers/clients really need, then select MVF(Most Valuable Features) that you can develop with good quality in the short term. Use such iterations to deploy quality updates, and not waste time and resources on unreasonable desires and sacrifice with quality.
11. Automation VS manual.
Automation is a 100% success in the long term. So if you have resources to automate something right now it should be done. You may think “it takes only 5 minutes, why I should automate this?”. But let's calculate this. For example, it’s an everyday task for a team of 5 developers. 5mins * 5devs * 21days * 12 month = 6 300mins = 105hours = 13.125days ~ 5250$. and how much this will cost if you have 40 000 employees?
12. Go out, get hobbies.
Work differentiation increases mental abilities and gives new fresh ideas. So make pauses and go out on fresh air, talk with friends, play the guitar, etc.