#feature toggle

Imagine you have a building. Some spaces are open and interconnected and anyone can move freely between them. Other spaces have doorways between them that restrict access. You can open and close the doors to control who goes where. Some doors require a key to get in.  Feature flags are like doors in a building. It would be wasteful clutter to put them everywhere, but when they’re placed strategically and when they’re well designed, they can be helpful.

Pros

Dark deploys: push code without users seeing it

Partial rollouts: beta test risky things

A/B testing: try things and see what works

Easy rollback: did something break? Just turn it off! 

Easier merges: feature flags > long running branches 

User segmentation: custom feature sets for different users

Feature dashboards: let stake holders decide when to turn something on

Cons 

Extra process: get ready to add 2-4 toggle related tasks to your to-do list

Tech debt: the longer your toggles live in the code the more they cost

Ugly code: feature toggles always feel messy 

Non-core value: people go to the circus to see the trapeze, not to see the safety net

Easy to misuse and abuse: toggles are like beer. The first few taste good. You’ll probably regret the tenth

Leaky features: toggle logic is code that can contain bugs just like any other code

Exponential testing: if you have features A, B & C, how many combinations are there?