Last night I was experimenting with sort.  It didn't take me long to notice that it wasn't sorting properly.  It's a specific failure mode, whereby the array has effectively been cut in two and then pieced together the other way round.

Having looked at internal\qsort.d and finding that's not the problem, I then went into the TypeInfo stuff and found this:

    int compare(void *p1, void *p2)
    {
        return *cast(int *)p1 - *cast(int *)p2;
    }

No check for overflow.  So indeed, sort has no reason to break the circle at int.min rather than any other value.

The fix is straightforward: do it in the same way as for longs.

For int:

    int compare(void* p1, void* p2)
    {
        if (*cast(int*) p1 < *cast(int*) p2)
            return -1;
        else if (*cast(int*) p1 > *cast(int*) p2)
            return 1;
        return 0;
    }

For uint:

    int compare(void* p1, void* p2)
    {
        if (*cast(uint*) p1 < *cast(uint*) p2)
            return -1;
        else if (*cast(uint*) p1 > *cast(uint*) p2)
            return 1;
        return 0;
    }

Stewart.

-- 
My e-mail is valid but not my primary mailbox, aside from its being the unfortunate victim of intensive mail-bombing at the moment.  Please keep replies on the 'group where everyone may benefit.