Re: [PATCH] honor TIME_STYLE environment variable for formatting ls timestamp

upon further reflection and experimentation, it seems like the easier 
solution to the referenced issue is the following (instead of a code patch):

env TIME_STYLE='+%s' find ~/Documents -exec ls -ld \{\} \+

one could alternatively invoke the ls command's parameter:

find ~/Documents -exec ls --time-style='+%s' -ld \{\} \+

and the invocation of ls directly will of course already include any 
predefined format-strings:

find ~/Documents -exec ls --time-style=full-iso -ld \{\} \+


I do think it might be handy to have a parameter to coërce the -ls time 
output from find into something sortable, like

find . -type f -ls -epoch

without involving the user having to write any explicit formatting 
strings at all; but this probably isn't a common enough usecase to 
justify introducing an entirely new parameter to find. Maybe an update 
to the documentation for -ls, though?  suggesting -exec ls if the full 
features of the ls command are necessary?

-Matt

On 11/15/25 13:05, Bernhard Voelker wrote:
> On 11/15/25 17:33, James Youngman wrote:
>> I think if someone requires a specific output format, then they should
>> specify the format they want with (for example) -printtf.
>
> excellent point - how could I have forgotten about -printf?
>> If the capabilities of -printf are insufficient to reproduce the
>> effect of, say, -ls, then we should improve it[*].   This general
>> problem is why the -printf documentation says:
>>
>>                %{ %[ %(
>>                       Reserved for future use.
>>
>> See commit 9cb1805e619b0f3aebd0984856de24964218bce2 of 2011-06-2011.
>>
>> If we're going to use one or more of these reserved sequences, I would
>> like one of them to be for user-specified extensions (since that's the
>> kind of use I had in mind when I reserved them).
>
> yes, one of the nice things about the --time-style option in ls(1) and
> du(1) are the predefined styles like "full-iso".
> Those could eventually be achieved with "%{ %[ %(", but the problem
> is that the time-style tells how to format, but now which of the
> timestamps: atime, btime, ctime, mtime.
> Maybe something like %{mtime:full-iso} but this also doesn't look ideal.
> There's many possibilities, even e.g. -ls-full-iso or alike.
> FWIW: there was a -lsh suggestion to print as in 'ls -dilsh' in human
> readable format a while ago.
>
> Anyway, I think with the above argument about the existing functionality
> in -printf, which would be redundant to the current time-style 
> suggestion,
> we can close the issue, right?
>   https://savannah.gnu.org/bugs/index.php?66299
>
> Have a nice day,
> Berny