It is with dismay that I sometimes find the trend of moving traditional configuration heavy programming models to convention based troubling. Case in point - I've been trying to adopt the newfangled Spring Data based repository framework. It's been working wonders and I got the QueryDSL support and naming convention based query generation going.
After creating a bunch of these repositories, I ran into an issue with one of my custom repositories. It was not recognizing that a method in the custom repo is part of the custom side and not to be a candidate method for spring to do its voodoo magic. For some reason Spring thought the method is a property of the entity bound to the repository. After digging through the Spring code base and spending a lot of time, it finally was clear why this was happening: I had made a typo on the repository naming and the "y" was missing from custom RepositoryImpl class. If only Spring was smart enough to tell me that this is what was causing the problem instead of point to a confusing message. So long type safety and compile-type checking, and welcome to the brave new world of magic based frameworks!
After creating a bunch of these repositories, I ran into an issue with one of my custom repositories. It was not recognizing that a method in the custom repo is part of the custom side and not to be a candidate method for spring to do its voodoo magic. For some reason Spring thought the method is a property of the entity bound to the repository. After digging through the Spring code base and spending a lot of time, it finally was clear why this was happening: I had made a typo on the repository naming and the "y" was missing from custom RepositoryImpl class. If only Spring was smart enough to tell me that this is what was causing the problem instead of point to a confusing message. So long type safety and compile-type checking, and welcome to the brave new world of magic based frameworks!
Comments