Bug 905 – printf strangeness

$ cat pp.d int main(char[][] args) { int i; ulong addr; for (i = 0; i < 10; i++) { addr = cast(ulong)(&i) + i; addr = ((addr + 3) >> 2) << 2; printf("%d %d %d \n", i, addr, i); } return 0; } $ dmd.exe pp.d g:\project\dmd\bin\..\..\dm\bin\link.exe pp,,,user32+kernel32/noi; $ ./pp.exe 0 1244976 0 1 1244980 0 2 1244980 0 3 1244980 0 4 1244980 0 5 1244984 0 6 1244984 0 7 1244984 0 8 1244984 0 9 1244988 0 (Note the 3rd column, all 0!)

> for (i = 0; i < 10; i++) { > addr = cast(ulong)(&i) + i; > addr = ((addr + 3) >> 2) << 2; > printf("%d %d %d \n", i, addr, i); > } Can you try this? printf("%d %lld %d \n", i, addr, i); the 'll' tells printf about the 'long' type. if you want to print a pointer type, use %p.

Not a bug, a ulong needs to be printed with %llu, not %d.

In general, you probably shouldn't be using printf unless there's a strong need for it. Use the more typesafe functionality provided by phobos with writef.

"Hello World" from DMD 1.0 uses printf. We'll be stuck with it for ages.