[Fwd'd email from Donn Terry]
(Yeah, I'm still lurking out here (for those that remember me)).
The -k option was introduced in the early 90s by the POSIX committee (with much
wailing and gnashing of teeth) because it would have been impossible to get the
standard approved without it. The output format required without -k was the
de-facto standard of System 5 at the time, and <that company> wasn't going to
accept something that invalidated their implementation. "Everybody knew" that
the -k format was the future (and when I use df, my fingers add -k), but...
The claim that the "de-facto standard is" is slightly specious: the "de-facto
standard appears to have become over time" is closer. However, since I have no
statistics on installed base and uptake rate, I couldn't say what the de-facto
standard actually is.
That said, it wouldn't be unreasonable to make such a change, but it should be
seen not as an error in the document so much as a conscious recognition that
the world has changed and it's time to change to match, which implies that a
definitive knowledge of how much the world has actually changed should be used.
I bet that TOG has the sorts of numbers that can be used to guide such a
decision.
Just my personal opinion, informed by ancient history.
Donn
-----Original Message-----
From: Roger Marquis [mailto:marquis@xxxxxx]
Sent: Sunday, November 08, 2009 9:09 AM
To: austin-group-l@xxxxxx
Subject: Re: df -k and free inodes
Since none of the most common implementation of 'df' return inode
information without '-i' perhaps it would be more productive to
conform POSIX to standard practice.
Standard practice is what systems admins expect, what their scripts
reflect, and what POSIX should standardize unless there is a compelling
reason to change differ.
The default behavior of 'df', whether including '-k' or not is:
Filesystem 1K-blocks Used Available Use% MountPoint
Neither 512K blocks nor inodes are returned. This is because that
information is rarely what users of this utility want.
Roger Marquis
> ======================================================================
> http://austingroupbugs.net/view.php?id=156
> ======================================================================
> Reported By: geoffclare
> Assigned To: ajosey
> ======================================================================
> Project: 1003.1(2008)/Issue 7
> Issue ID: 156
> Category: Shell and Utilities
> Type: Error
> Severity: Objection
> Priority: normal
> Status: Interpretation Required
> Name: Geoff Clare
> Organization: The Open Group
> User Reference:
> Section: df
> Page Number: 2595
> Line Number: 83575
> Interp Status: Proposed
> Final Accepted Text: http://austingroupbugs.net/view.php?id=156#c289
> ======================================================================
|