int-->float

[Wendy Thrash]

Doug McIlroy rightly calls the float-int comparison example we've been
discussing a "bug" rather than a bug.  Unfortunately, he then proposes to
fix this "bug" by automatic widening.  Well, when I cast an integer to
float, I do it for a very specific reason, and I really prefer not to
have the language/compiler circumvent my purpose.  If people want to
convert to the widest floating type available, let them cast to _that_,
for heaven's sake.  Don't destroy my old code in an attempt to make someone
else's pseudo-portable code really portable.

-Wendy T., who's been bitching about these conversions for years,
           and likes the direction they've been taking.