Re: df command should suppress duplicates

[Ondrej Oprala]

On 11/29/2012 06:45 PM, Bernhard Voelker wrote:
> On 10/04/2012 07:36 PM, Ondrej Oprala wrote:
> > Hi, I haven't heard from this thread in a while, are there still things
> > to be done here?
> Hi Ondrej,
>
> FWIW, I've rebased your patch to Git master.
> (The only change I made - besides resolving the conflicts -
> was to declare devlist_head static to make syntax-check happy.)
>
> However, that implementation needs further discussion:
>
> * IMO, "df -a" should not skip bind mounts, and instead
>    print all of the lines.
>
> * It doesn't play well with Robert's suggestion for hiding
>    the 'rootfs' entry, because dev_examined() already added
>    the entry for the '/' file system before the rootfs entry is
>    skipped, which makes the real '/' entry also be skipped.
>    http://lists.gnu.org/archive/html/coreutils/2012-11/msg00091.html
>
>    But also Robert's patch deserves further discussion.
>    IMO e.g. "df -t rootfs" and again "df -a" should not skip
>    rootfs. And read_file_system_list must be called with true
>    param to make it work.
>
> * It still allocates one devlist entry more than needed.
>
> Have a nice day,
> Berny
Hi Bernhard,
thanks for the rebase :) .
I've modified the patch a bit, so I dont interfere with output if df -a 
is specified.

Regarding Robert's suggestion/patch and your opinions on it,
I agree that df -t rootfs should indeed output info about it.
So rootfs is now filtered as well as usual duplicates by default,
but should show up with the -[at] options.

As for the extra entry...sorry but I don't see it, the first entry 
(allocated
in get_all_entries) is filled/used as well as all the others.

Thanks,
Ondrej.

DIFF
Description: Text document