- There are several different
- Think about collaborating on these operations
- Thinking about adding [Common Interfaces
- Each separately Cost to learn is a waste of money.
- Different plurals have different areas of expertise.
- Ideally, both sides should be able to take advantage, but often both sides only have common areas and lose one’s strength.
If A has a function that B does not
-
Route 1: Unsupported routes
- Implementation is easy
- It’s the unfortunate thing I wrote above.
-
Route 2: Route to error
- For example, a design that has an interface but causes an error when executed
- In the end, the user must understand “what is available on which backend”.
-
Line 3: The line that you do your best to implement
- Example: The machine learning library Scikit-Learn provides a Common Interface for different learning models.
- For model m, method
m.predict_probareturns the result of probability estimation - This method is also grown for SVMs that do not have the capability of probability estimation
- Running SVM 5 times logistic regression on 4/5 data: Probability Estimation with SVM.
- For model m, method
- Very labor intensive.
- Not always possible to implement.
- In that case, it is attributed to route 2.
- Example: The machine learning library Scikit-Learn provides a Common Interface for different learning models.
-
snare in Common Interface #Software Design #Design
This page is auto-translated from /nishio/共通インターフェイスの罠 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.