The initial impetus for creating this library was entirely practical. I had been working on another library of macro functionality, which used Boost PP functionality, and I realized that if I could use variadic macros with my other library, the end-user usability for that library would be easier. Therefore the initial main design goal of this library was to interoperate variadic macro data with Boost PP in the easiest and clearest way possible.
This led to the original versions of the library as an impetus for adding variadic macro data support to Boost PP. While this was being done, but the variadic macro data support had not yet been finalized in Boost PP, I still maintained the library in two modes, either its own variadic data functionality or deferring to the implementation of variadic macros in the Boost PP library.
Once support for variadic data had been added to Boost PP I stripped down the functionality of this library to only include variadic macro support for functionality which was an adjunct to the support in Boost PP. This functionality might be seen as experimental, since it largely relied on a macro which tested for empty input which Paul Mensonides, the author of Boost PP, had published on the Internet, and which by the very nature of the C++ preprocessor is slightly flawed but which was the closest approximation of such functionality which I believed could be made. I had to tweak this macro somewhat for the Visual C++ preprocessor, whose conformance to the C++ standard for macro processing is notably incorrect in a number of areas. But I still felt this functionality could be used in select situations and might be useful to others. Using this functionality I was able to build up some other macros which tested for the various Boost PP data types. I also was able to add in functionality, based on Paul Mendsonides excellent work, for handling tuples in preprocessing data.
All of this particular functionality is impossible to do effectively without the use of variadic macros. But I had kept these features at a minimum because of the difficulty of using variadic macros with compilers, most notably Visual C++, whose implementation of variadic macros is substandard and therefore very difficult to get to work correctly when variadic macros must be used.
I then realized that if I am going to have a library which takes advantage of variadic macros I should see what I could do in the area of parsing preprocessor data. This has led to a reorganization of the library as a set of macros largely for parsing preprocessor data. All of this is now built on top of my use of the almost perfect checking for emptiness which Paul Mensonides originally created.