bash: any way to reuse the last output?

[Tom Mitchell]

On Fri, Jan 23, 2004 at 03:20:17AM -0200, Herculano de Lima Einloft Neto wrote:
> 
> Hello all,
> 
>    This happens all the time: after an intricate find, I decide
> to grep something out of that very same find, for instance. It
> would be nice to have some tweak to reuse the last output without
> having to run the command again. Anyone?

It all depends on what you are tweaking.

If the first "find" contains all the data you are looking for just
save the output to a file.  Since this "happens all the time" get in
the habit and use a scratch file.

Something like:
In window one:
  find / -xdev -type f -name foo -o this -o that > /tmp/find.out
  wait; tput bel; sleep 1;tput bel; sleep 1;tput bel; sleep 1;tput bel
In window two:
  tail -f /tmp/find.out | less

There is an obvious caution, some searches can generate a gazillion
bytes of output filling up the file system.  You also have to know in
advance that you are going to tweak your "find".  This is the biggest
reason that there is no 'standard' tool for this.  For me the output
of a command three back is really the one I want to tweak.  So command
history is the big win.

N.B. "find" has options that are rich in file system meta data such as:
	-user -mtime -size -newer ... etc.
If you tweak these you have to rerun the "find" because this info is
not in the tmp file.  Example

	touch /tmp/then
	logger nap-time
	sleep 20
	logger nap-over
	touch /tmp/now
	find / -xdev -newer /tmp/then > /tmp/then.out
	find / -xdev -newer /tmp/now  > /tmp/now.out

Even though /tmp/then.out contains all the file names that you might
want in /tmp/now.out the meta data is missing that would permit you to
process the first file into the second.

The good news is that the first meta data rich find can load the file
system meta data into DRAM so the second might be quicker (if there is
enough DRAM).  If you do cache busting stuff all bets are off. Thus:

	find / -xdev -type f -print0 | xargs -0 grep magicword

One tool I find very valuable is slocate/locate. It keeps a data base
(cache) of file names that I find very handy. It is much quicker than a
simple find looking for file names.  I use "locate" a LOT.
    $ rpm -q --whatprovides `which locate`
    slocate-2.7-4
    $ ls /etc/cron.daily/slocate.cron

-- 
	T o m  M i t c h e l l 
	mitch48-at-sbcglobal-dot-net