As there is an increasing number of C++ versions in production use, the challenge of making all of one's object files and libraries ABI compatible has become more urgent. It has always been important given the C++ build model, but increasingly users run into actual errors caused by this. This talk will explain how library developers can make ABI compatibility mistakes less likely to cause silent memory corruption, and more likely to cause link errors. Additionally it will talk about how library users can help. It will also touch on the importance of having build systems and tool chains that are aware of ABI compatibility.
