What about Performance?

[qoomon]

Hi, I really like your mapping library and its api. Are there any performance benchmarks comparing manually mapping vs mapstruct and shapeshift?

[Idane]

Generally we've not done any comprehensive benchmarks yet, but performance wise ShapeShift will most likely fare worse to mapstruct, seeing as they use code generation and have very little overhead at runtime whereas we made the choice to map at runtime for several reasons;

1. It centralizes the mapper. (No need for XToYMapper, everything can be mapped via ShapeShift by just supplying the source and target)
2. It doesn't require the user to add an annotation processor to their project.
3. In addition to mapping by annotations, we also support a Kotlin DSL for defining mappings. As the DSL is parsed at runtime, there is no way for us to generate code from it without further complicating the development experience for the end user.
4. We also put a heavy emphasis on Spring use and transformer beans, the ability to have context-aware mappings which may make additional calls (For example, to a database) or checks with other beans is crucial to our design.

Leaving this open so we can address this, thanks!

[nguchakov]

It would be great to be mentioned in the documentation. Somewhere at overview section. Tried to find before found this issue