Actually, I was thinking more of existing libraries. Case in point. I get relatively little feedback on the boost serialization library. It's been 20+ years. Initially it was almost alone. But now there are lots of alternatives. Serialization has never been proposed for the standard. It has few "stars" on github. I'm wondering if it's even used any more. It's becoming increasingly out of sync with evolving boost tools and hence more effort to maintain. Perhaps it's time to deprecated it or remove it from boost.
Hello Robert, I can confirm that, here in our production code since 2005, we are intensively using your Boost serialization library. It was very well designed and its interoperability with other Boost libraries (utilities, multi_index, bimap...) is very valuable. From my point of view, its deprecation will be really a mess. I am very sorry you have that feeling. Best regards, -- Marc Viala