Re: REQ: -g for --block-size=1G

[Eric Blake]

> I looked through the archives and didn't see anyone ask for this
> before, so I will.  I would love for 'df -g' and 'du -g' to act like
> the -m option and set the block size.

Whether or not we add -g, we have some work to do to
make things consistent with -m.

In du:
-m is documented, and means --block-size=1M
--megabytes is undocumented, prints a deprecation
warning, then falls through to -m

In df:
-m and --megabytes are undocumented, and mean
--block-size=1M without warning

However, the reason that --megabytes was deprecated
in du also applies to df - namely, 1M stands for mebibytes
(1024*1024), while 1MB stands for megabytes (1000*1000).

At least our general policy that every short option should
have a corresponding long option is met, since -m matches
--block-size=1M, even though the long option doesn't
start with m.

Now, with regards to your request about -g - we are very reluctant
to add short options without good reason.  If there is another
implementation out there using the same letter for the same purpose,
then that is a good reason.  But what's wrong with continuing to use
the long option --blocks=1G (or, so long as we don't introduce any
ambiguous long options, the equivalent --b=G)?

-- 
Eric Blake

-- 
View this message in context: 
http://www.nabble.com/REQ%3A--g-for---block-size%3D1G-tp20861569p20864739.html
Sent from the Gnu - Coreutils - Discuss mailing list archive at Nabble.com.