Possibly unintended change to the sorting of hardlinks about qdirstat HOT 1 CLOSED

It becomes clearer when we see a bit more context:

So we have a lot of small files that all consume one cluster (4k), so the other rule comes into play:

https://github.com/shundhammer/qdirstat/blob/master/src/FileInfoSorter.cpp#L40-L47

if ( a->totalAllocatedSize() == b->totalAllocatedSize() )
{
    // This is a common case because of cluster-wise allocation
    return a->totalSize() < b->totalSize();
}
else
    return a->totalAllocatedSize() < b->totalAllocatedSize();

All those tiny files are sorted by their totalSize(), the byte size because their totalAllocatedSize() is the same every time: 4k, so you see them sorted by the few bytes of their nominal size.

But when compared with the files with multiple hard links, their totalSize() is no longer the same: Because the files with only one hard link still have one cluster (4k), while those with multiple hard links share that one cluster between all their hard links, so it's actually less than that one cluster; in this example one cluster divided by their two links, i.e. 4k / 2 = 2k.

That might be a bit irritating because it looks like there is a jump in the sorting, but it's really consistent.

from qdirstat.