Le Dat’s Reviews > The Pragmatic Programmer: From Journeyman to Master > Status Update

On estimation, remember giving both optimistic and pessimistic scenario.
Ex9. If a man could walk between computers and inject the 1TB storage, OS recognized and automatically transferred data within a sec, then it is faster than a 1 Gbps bandwidth.
Ex10. the above “dream” claims that the man is faster. In deed, we can made a formula to weight out bandwidth in reality: storage * velocity * cp speed vs. 1 Gbps

DSL (DL) involved me once when we wrote a parser to translate text into test for integration. It was funny but it had too many devils. Now challenges:
- yes, can. The idea of requirement expression compiling into code can be real, but it has to be pure and I dont think real world reqs can be pure
- it has to be detailed on purposes, eg. rails- is solving a web development problem, well-documentation, open-sourced

A tracer bullet metaphor is
- for engineering: likely a BDD approach when bullet is a test to ensure output is guaranteed
- for product: a vision of product discovery to get users or stakeholders see something they always want to

But it is not a prototype, it will never hit the target that hard. We are just adding brightness.

The Split2 is more clear, orthogonal and simple as it just receive a line as an input. We are not getting much side effects due to outside context.

Speaking of OOP and FP, OOP has more dependencies wrapped in its instance that you have control on not only instance/object but also its inheritance/mixins. FP paradigm usually passes the context as param then it is more clear.

But structuring code on detail is a creep.

At the philosophy of Decorator, Singleton or Strategy Pattern to cherry pick when avoid coupling or similar functions