[Bug 92] New: std.format %x, %o and %b assume 64-bit negative values

[d-bugmail at puremagic.com]

http://d.puremagic.com/bugzilla/show_bug.cgi?id=92

           Summary: std.format %x, %o and %b assume 64-bit negative values
           Product: D
           Version: 0.152
          Platform: PC
        OS/Version: Windows
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Phobos
        AssignedTo: bugzilla at digitalmars.com
        ReportedBy: smjg at iname.com


When the %x, %o or %b format is used to format a negative integer, the correct
result is produced only in the case of long.  Even for byte, short or int, the
output is 64 bits long.

----------
import std.stdio;

const char[][] fmts = [ "%x", "%o", "%b" ];

void main() {
    foreach (fmt; fmts) {
        byte b = byte.max;

        writefln(fmt, b);
        writefln(fmt, ++b);
        writefln(fmt, ++b);

        short s = short.max;

        writefln(fmt, s);
        writefln(fmt, ++s);
        writefln(fmt, ++s);

        int i = int.max;

        writefln(fmt, i);
        writefln(fmt, ++i);
        writefln(fmt, ++i);

        long l = long.max;

        writefln(fmt, l);
        writefln(fmt, ++l);
        writefln(fmt, ++l);

        writefln();
    }
}
----------

The %x case should suffice to illustrate.
Actual output:

7f
ffffffffffffff80
ffffffffffffff81
7fff
ffffffffffff8000
ffffffffffff8001
7fffffff
ffffffff80000000
ffffffff80000001
7fffffffffffffff
8000000000000000
8000000000000001

Expected output:
7f
80
81
7fff
8000
8001
7fffffff
80000000
80000001
7fffffffffffffff
8000000000000000
8000000000000001

The same happens if I specify e.g. %04x instead of %x.


--