Package: tech-ctte
Severity: normal
X-Debbugs-Cc: 761...@bugs.debian.org

As reported in #761424, the "sizes" output by df(1) are not actual sizes, due to the 
"units" used. The "units" used are merely unit prefixes. Moreover, df uses the same 
pseudo-units with different meanings. The output can therefore be misleading if one attempts to interpret it. 
2 different calls seem to give different measurements for the same filesystems:
chealer@debian:~$ LANG=C df -h
Filesystem                                              Size  Used Avail Use% 
Mounted on
rootfs                                                  1.8T 317G  1.4T  19% /
udev                                                     10M 0   10M   0% /dev
tmpfs                                                   252M 5.4M  247M   3% 
/run
/dev/disk/by-uuid/a00f8767-e954-4b81-8035-c6bb414671cb  1.8T 317G  1.4T  19% /
tmpfs                                                   5.0M 4.0K  5.0M   1% 
/run/lock
tmpfs                                                   961M 0  961M   0% 
/run/shm
/dev/sdb2                                               2.2G 122M  2.0G   6% 
/tmp
chealer@debian:~$ LANG=C df -H
Filesystem                                              Size  Used Avail Use% 
Mounted on
rootfs                                                  2.0T 340G  1.6T  19% /
udev                                                     11M 0   11M   0% /dev
tmpfs                                                   264M 5.7M  259M   3% 
/run
/dev/disk/by-uuid/a00f8767-e954-4b81-8035-c6bb414671cb  2.0T 340G  1.6T  19% /
tmpfs                                                   5.3M 4.1k  5.3M   1% 
/run/lock
tmpfs                                                   1.1G 0  1.1G   0% 
/run/shm
/dev/sdb2                                               2.4G 128M  2.2G   6% 
/tmp
chealer@debian:~$

In the first example call, JEDEC prefixes are used, while SI prefixes are used 
in the second.

Michael Stone denies, excusing the behavior with space scarcity, documentation, 
and what he considers as little impact:
I'm not going to deviate from upstream. [...] The
space is more important (in my opinion) than the need for a constant
reminder of the unit. The documentation is there for people to read the
first time, after that it's just not that important. (Even for the
numbers above the difference isn't really significant--the relative
sizes are the consistent, and what are the odds that you need exactly 12
gigasomethings? If you did need exactly that much space, you're probably
better off looking at kbytes or bytes anyway.)

The space scarcity and impact arguments do not hold, and unfortunately, even 
the documentation does not define the pseudo-sizes currently output.

--
Filipus Klutiero
http://www.philippecloutier.com

Reply via email to