Of course not.

$ perl -MDevel::Peek -E 'Dump $_ for 1<<63,-(1<<63)'
SV = IV(0x559e408d8210) at 0x559e408d8220
  REFCNT = 1
  FLAGS = (IOK,pIOK,IsUV)
  UV = 9223372036854775808
SV = IV(0x559e408d83c0) at 0x559e408d83d0
  REFCNT = 1
  FLAGS = (IOK,pIOK)
  IV = -9223372036854775808

$ perl -MJudy::L=Set -E 'say Set($j, $_, 1) for 1<<63,-(1<<63),0'
94434583711008
Coercing -9223372036854775808 to 0. Can't use negative values as keys.
+ at -e line 1.
94434579767256
94434579767256
[download]

You didn't retrieve any information from the data structure, so you didn't test it at all. Try this:

perl -Mv5.14  -MJudy::L=Get,Set -e'
   my $i = 1<<63;
   my $j = -$i;

   my $judy;

   Set( $judy, $i, 4 );
   Set( $judy, $j, 5 );

   say( ( Get( $judy, $i ) )[1] );
   say( ( Get( $judy, $j ) )[1] );
'
[download]

If you get different values, I was wrong. If you get the same value, each negative value is considered to the same as a very large positive value.

Without the hack, it outputs 5 twice. I suspect the hack only silences the warning.

Of course not.

No, it isn't obvious that the keys are 65 bits (or larger) in size. In fact, they pretty obviously aren't.