`Base.Printf.ini_dec` has a fall-thru issue

Some of my more precise floating point types do not properly know how to overload Base.Printf.ini_dec or any of its very close algorithmic partners Printf.fix_dec Printf.decode_dec (nor which of those should be given special handling). I understand that the abc_dec functions are intended to include fall-back implementations that handle or otherwise appropriately pass through a value-consistent result.

Here is some code that fails at the moment: https://github.com/JuliaMath/DoubleFloats.jl/issues/81

Something is not fully sewn on that side. I would like to know what I should provide within DoubleFloats that would allow the values to proceed through the Printf setup and construction correctly. And also what special Base-internal (not exported functionality would be helpful).

Posted my thoughts here

1 Like