Yet another iteration of the solution to the floating-point conversion
problem. This one avoids complaints about oversize double values
from -fsanitize. (tags: branch-3.47)
↧