As an optional usage in coded-arithmetic and hexadecimal constants, where it is unambiguous, this is an excellent idea.
It has been available in PL/I for a long time, where I may write Declare Fmax value(2_147_483_6747) signed binary fixed(31,0) ; or Declare Fmax signed binary fixed(31,0) static initial(2147483647) ; Declare Fmax initial(0111_1111_1111_1111_1111_1111_1111_1111b) static binary fixed(31,0) ; as may be contextually convenient. I have found it useful there. It reduces the frequency of trivial errors, improves readability, and in in general labor-saving. John Gilmore, Ashland, MA 01721 - USA
