codez

var headerParts = headerLine.Split(new[]{':'}, 2);

class Util {
  static string[] SplitHeader(string header) {
    return header.Split(new[]{':'}, 2);
  }
}

var headerParts = Util.SplitHeader(headerLine);

class StringHeaderUtil {
  static string[] SplitHeader(string header) {
    return header.Split(new[]{':'}, 2);
  }
}

var headerParts = HeaderUtil.SplitHeader(headerLine);

DRY

DRY

▲ === HeaderUtil

what happens when headerLine == "nocolon"?

1. new[] { "nocolon" }
2. new[] { "nocolon", "" }
3. new[] { "nocolon", null }

in A, if (headerParts.Length == 1)

in B, if (headerParts[1] == null) and change SplitHeader to return new[] { parts[0], null }

Acyclic Dependencies Principle is ☹

reuse => dependency and coupling

need to find the ballance

Uncle Bob's principles of OOD

Best Practices

"A best practice is a method or technique that has been generally accepted as superior to any alternatives"

Best Practices?

• Stored Procedures!

Best Practices?

• Stored Procedures!
• Waterfall
• SOAP
• REST

Best Practices are myth!

Bad Practices

• while(*dst++=*src++);
• give interns unlimited admin access to prod db
• use your phone while biking

Worst Practices

• not proactively practicing disaster recovery
• shut your eyes while biking
• reversible passwords

practices

Good Bad Worst Best

TDD

RED => GREEN => REFACTOR

is NOT a design tool

You see, it is not TDD that creates bad designs. It is not TDD that creates good designs. It’s you. TDD is a discipline. It’s a way to organize your work.

DHH: Test induced design damage

Uncle Bob: bad designers induce design damage

Ken: TDD applied badly often lead to design damage

sub-units, integration classes, excessive use of mocks, isolating stable code

public void IsPrime_One() {
	Assert.True(Primes.IsPrime(1));
}

bool IsPrime(int number) {
	return number == 1;
}

switch (number) {
  case 1: 
  case 2:
  case 3:
  case 5:
    return true;
    break;
  default:
    return false;
}

//algorithm

We are paid to deliver lasting business value
<| we write high quality maintainable code
<| we might write automated tests
<| we aim at high coverage
<| we can choose TDD as a discipline, or not

How many?

How many?

Teams? a whole lot

Services? a whole lot

Network calls? a whole lot

Resources spent de/serializing? a whole fucking lot

Microservices and SOA at Amazon

Microservices and SOA at Amazon

Problems

• Latency
• Access exceptions
• Cost
• Service discovery
• Consistency
• (imagine the 'everything is fine' gif)
• Friendly-fire DoS attacks
• Debugging
• Pager escalations

Find the differences

basically trading 💩 for a whole bunch of 💩💩💩

Service boundaries are logical and can be physical

Decoupling is key