Re: [PATCH] Support for --size in du

[Sami Kerola]

On Sun, Jan 20, 2013 at 10:55 AM, Pádraig Brady <address@hidden> wrote:
> On 01/19/2013 10:47 PM, Bernhard Voelker wrote:
>>
>> On 01/17/2013 03:58 PM, Pádraig Brady wrote:
>>>
>>> I Just noticed another option from BSD:
>>>
>>> -t threshold
>>>        Display only entries for which size exceeds threshold.  If
>>>        threshold is negative, display only entries for which size is
>>>        less than the absolute value of threshold.
>>>
>>> I slightly prefer --min-size and --max-size,
>>> though not at the expense of losing the opportunity
>>> to increase compat between these tools.
>>>
>>> I suppose we could rename --size to --threshold too.
>>
>>
>> It seems there are many ways - like always. ;-)
>>
>> I think --min-size/--max-size has a functional plus over one
>> single new option. But increasing compatibility to BSD is tempting.
>>
>> If I'd be asked to decide between --size and --threshold then
>> the latter is the winner: a) BSD compat, b) the abbreviation
>> of the option better fits into the already existing ones,
>> see --size vs. --si, while --th is (currently) fine.
>>
>> I'll change the patch towards --threshold then. WDYT?
>
>
> +1
>
> I don't think adding extra --{min,max}-size options are worth it.

My 2 pennies, might be off/new topic.

I am often looking for greatest N files/directories in size. For
instance when partition is nearly full then the line I write is

du -max /var/ | sort -n | tail -10

In my fantasy world there would be new long only option in du doing the same.

du -hax --top=10  # the 10 could be default when optional option is not defined.

That would answer to the question I often have in mind; what is
filling the disk?

I am not sure would I ever remember to use --min-size && --max-size du
options. The find command does excellent job for that purpose, and
it's burned into at least to my memory as a utility to do that job.
Another drawback with --max-size is that when I am interested what is
filling disk I need to guess a limit. If I get it wrong the listing is
either very long or nothing, and rerun is required. The guess
iteration loop can be avoided if --top is present.

-- 
   Sami Kerola
   http://www.iki.fi/kerolasa/