I feel that this is something that would be fixed by writing your test suites first. If you think you have designed a tremendously clever API, but then the code that calls it is an absolute mess, then you have not designed a tremendously clever API.
General principles of restraint and not having to use every available language feature all the time apply here as everywhere - keep it simple, stupid. The only WTF I want when reading your code is, “I didn’t realise there could be such a straightforward way to do it”.
I had designed a system with what I thought was a good API, wrote the test for it and I was satisfied how simple it was to use the thing but then, when it came to integrate it, it was clunky and no good way to actually use the system.
I don’t think it’s always obvious when an API is good or bad, even with tests.
I agree we should use operator overloading only when it really fits the use case. Especially the function call operator is easily misused.
I don’t completely agree with Raymond Chen here though. Firstly I don’t think providing both an explicit Load() function and the function call operator is the solution. Just keep things simple and obvious: provide Load() and remove the function call operator.
Secondly, why is StorageLoader even a class (or actually a struct here, but we know that’s the same thing in C++)? Unless Raymond is leaving out something essential, there is no state. Just make a function:
template<typename DataType> LoadFromStorage(StorageOptions<DataType> const* options) { // … }
I’ve been annoyed at this in Python sometimes, since its containers have some operations named only, some operator only, and some available as both.
C++'s standard containers mostly have useless operator overloads though, so its technical win here isn’t particularly impressive. One thing you definitely notice when porting C++ code to other languages is just how many silly dances various APIs force you to use for common tasks.
I agree we should use operator overloading only when it really fits the use case. Especially the function call operator is easily misused.
I don’t completely agree with Raymond Chen here though. Firstly I don’t think providing both an explicit Load() function and the function call operator is the solution. Just keep things simple and obvious: provide Load() and remove the function call operator.
Secondly, why is StorageLoader even a class (or actually a struct here, but we know that’s the same thing in C++)? Unless Raymond is leaving out something essential, there is no state. Just make a function:
template<typename DataType> LoadFromStorage(StorageOptions<DataType> const* options) { // ... }I agree we should use operator overloading only when it really fits the use case. Especially the function call operator is easily misused.
I don’t completely agree with Raymond Chen here though. Firstly I don’t think providing both an explicit Load() function and the function call operator is the solution. Just keep things simple and obvious: provide Load() and remove the function call operator.
Secondly, why is StorageLoader even a class (or actually a struct here, but we know that’s the same thing in C++)? Unless Raymond is leaving out something essential, there is no state. Just make a function:
template<typename DataType> LoadFromStorage(StorageOptions<DataType> const* options) { // ... }This solves all the problems: you can simply call it, even without operator overloading (because it’s already a function), and doesn’t make it awkward to specify the data type we need. We’re writing C++ here, not Java where everything has to be a class even if it doesn’t have any reason to be.
