On 13:59:09, 29.02.16, Theo de Raadt wrote:
> > On 2016-02-29, Michal Mazurek <[email protected]> wrote:
> >
> > > If the number of filesystem inodes or blocks is too big, the columns in
> > > the
> > > output of df will become misaligned.
> > >
> > > I'm resending the patch that adds variable width columns to df.
> >
> > I am somewhat worried about scripts that examine the df output and
> > expect the columns to be in fixed places.
>
> As am I. Back in the day, that practice was very common.
I could imagine that, but the columns aren't in a fixed place if the number
of blocks or inodes is greater than a particular value, wouldn't this break
those scripts already?
And if you meant comparing the header as a string, then the first column
isn't fixed either, it depends on the length of the longest disk path:
(void)printf("%-*.*s %s Used Avail Capacity",
maxwidth, maxwidth, "Filesystem", header);
I wanted to make similar changes to ps(1), the TIME and VSZ/RSS columns
can misalign the table. I suppose the same argument can be made against
that change.
--
Michal Mazurek
