picostitch
crafting (and) JavaScript
#architecture

Technical Debt is Anything Preventing You From Developing Fast

A podcast with @Sandro Mancuso talking about tech debt and maintainable software. My notes.

What is Maintainable Software?

  1. "Tested in minutes, if not in seconds"
  • "I am not scared to change"
  • "Press a button run all the tests and see it works"
  • "I can reply on the test suites"
  1. "A change is localized"
  • make a change in one place and I dont need to change anything else
  1. "Find the place to change"
  • how the software is designed
  • the language used should align with the business, with what the software does
  1. "Relationship between modules" (module being function, class, …)
  • coupling + cohesion
  • Fun in: the number of modules talking to this module
  • Fun out: "one module talks to too many modules"
  1. "Who is making the change" the human side
  • OO or functional
  • "what you are used to and what you know"

DDD in Software

"When you look at the package structure, how do you know what the package does?" (package = namespaces)

  • only tech layers hide what the software does
  • classes are nouns, but is a "User" an "Author", "Buyer", …?
  • what do we do with those
  • "The behaviour of the system is very rarely seen in the package and the classes" "very rarely seen this"

How do you define Tech Debt?

"Anything the system is preventing the system to achieve"

  • even badly written code that needs NO change, its no tech debt
  • tech debt is what is preventing us to be fast
  • devs sometimes call things they don’t like tech debt
  • "every tech problem is a business problem" and vice versa

Structure of the Team

  • "POs prioritise a back log and then go to devs" this misses including the devs
  • the business does not understand refactoring and plays it back, we (devs) need to provide a strategy and why - you get a "no"
  • prevent the disconnect between biz and tech

Outside in Development

Frontend First Dev, Interaction Driven Design

  • "There is no point in having a backend if there is no frontend"
  • "we design our domain first and the persistence and frontend is just a detail"
  • "we should analyze how we interact with the system, that’s where I would start from"
  • "understand what the external world wants from my system"
  • growing a domain model is like TDD from very specific tests which drive generic code, the system can evolve this way too

(Software) Design

There is a practical and a theoretical side.

  • "you can not design software well by just writing code"