开云体育

?

I find this pretty provocative, so of course, that interests me immediately. :)

Two questions leap to mind:

- What kind of "weak" is your "weak" here?

- Which "survey literature" has drawn conclusions about the strength of design error signals? Most of the attempts I've seen have measured what happens when people who don't know much about design try to write tests first, notably undergraduate computer science students.

To be clear, I routinely experience design signals that are relatively weak individually, but become stronger in aggregate, such as "I see 3 reasons to do refactoring X, so now I feel confident that doing it will help". I also notice that, as I become more experienced, I react to the signals more softly, meaning as suggestions or hints, rather than as urgent calls to action. I don't consider that a weakening of the signal, but rather an indication that I have become less dogmatic in my practice.

To provide examples of how I think of "strong" or "weak", I can think of at least two common examples of design signals that I'd label "strong":?

- scattered duplication of behavior focused on a naked, primitive data type (a number or some text) which points towards introducing a Whole Value type

- rampant duplication in application request handlers which points towards introducing Parsers and extracting Standard Workflows

If you didn't know those options existed, I'd expect duplication in either production code or tests to provide the motivation to seek them out.