On Feb 15, 2018 2:30 PM, Paul Eggert <eggert@cs.ucla.edu> wrote:

On 02/15/2018 02:13 PM, Daniel Colascione wrote:
> I'm tempted to just convert INTEGER_TO_CONS to a pair of functions
> (one signed, one unsigned) that just take an {,u}intmax_t.

Sometimes one does not know whether an integer is signed. For example,
the ino_t type is signed on some platforms, and not on others. This is
why INTEGER_CONS is written the way it is: it is supposed to do the
right thing regardless of the signedness of its argument.


<heresy>C++ would solve this problem</heresy>