bug#44273: "total used in directory 19 available 5.2 GiB"

[Eli Zaretskii]

> From: Mattias Engdegård <mattiase@acm.org>
> Date: Sun, 1 Nov 2020 12:55:23 +0100
> Cc: Juri Linkov <juri@linkov.net>, Eli Zaretskii <eliz@gnu.org>, rms@gnu.org,
>         jidanni@jidanni.org, stefankangas@gmail.com, 44273@debbugs.gnu.org
> 
> 1 nov. 2020 kl. 12.38 skrev Lars Ingebrigtsen <larsi@gnus.org>:
> 
> > It only displays the space used directly in the current directory, which
> > is (in general) not interesting information.  (I.e., it doesn't descend
> > into sub-directories and count the usage there, so in any directory that
> > has a sub-directory, the data is misleading at best.)
> 
> Right, and it's worse. Even if we knew exactly the unit used (sectors, 
> kibibytes, imperial qubits), what we get for a subdirectory is still 
> platform-dependent.
> 
> Except that it's worse than that. The semantics may vary between file-systems 
> on a single machine.
> 
> Except that it's even worse. It may depend on the history of the 
> subdirectory. For instance, it could be the maximum number of disk hardware 
> sectors used for the internal representation of some of the metadata for that 
> directory at any time in the past. (This is probably the original meaning of 
> the number.)

This actually means that this value provides information that is not
directly available in the file sizes.  So it is useful.

I find the notion that we should remove from display a useful quantity
just because it might be tricky to interpret in some situation,
especially by people who are not proficient in this stuff.  'ls' shows
it, and Dired isn't supposed to lose stuff that 'ls -al' provides.