[9fans] fun with rc

[9fans] fun with rc

[erik quanstrom]

so an interesting problem i run into from time to time is separately computing
the files added to and deleted from a directory in a shell script.  uniq doesn't
work for this.  certainly one can loop over two lists of files and do this easily,
but that seems dull and tedious.

but as it turns out, one can compute the deleted and added files relatively
efficiently with diff in two steps.  obviously uniq -d gives us the union, so

	fn ∩ {echo $$1 $$2 | sed 's/ /\n/g' | sort | uniq -d}

and uniq -u gives us not union

	fn not∩ {echo $$1 $$2 | sed 's/ /\n/g' | sort | uniq -u}

but then we can compuete the "difference" (relative complement) we want
with

	fn listminus {c=`{∩ $1 $2} not∩ c $1}

so then

	echo 'added=('  `{listminus old new} ')'
	echo 'deleted=('  `{listminus new old} ')'

he!  such fun with rc.

this is available now in 9atom.

- erik

Re: [9fans] fun with rc

[minux]

[-- Attachment #1: Type: text/plain, Size: 1057 bytes --]

On Mar 12, 2015 12:52 PM, "erik quanstrom" <quanstro@quanstro.net> wrote:
> so an interesting problem i run into from time to time is separately
computing
> the files added to and deleted from a directory in a shell script.  uniq
doesn't
> work for this.  certainly one can loop over two lists of files and do
this easily,
> but that seems dull and tedious.
>
> but as it turns out, one can compute the deleted and added files
relatively
> efficiently with diff in two steps.  obviously uniq -d gives us the
union, so
>
>         fn ∩ {echo $$1 $$2 | sed 's/ /\n/g' | sort | uniq -d}

Isn't this called (set) intersection?

> and uniq -u gives us not union
>
>         fn not∩ {echo $$1 $$2 | sed 's/ /\n/g' | sort | uniq -u}

this is (set) symmetric difference.

> but then we can compuete the "difference" (relative complement) we want
> with
>
>         fn listminus {c=`{∩ $1 $2} not∩ c $1}
>
> so then
>
>         echo 'added=('  `{listminus old new} ')'
>         echo 'deleted=('  `{listminus new old} ')'
>

[-- Attachment #2: Type: text/html, Size: 1437 bytes --]

Re: [9fans] fun with rc

[erik quanstrom]

On Thu Mar 12 10:09:06 PDT 2015, minux.ma@gmail.com wrote:

> On Mar 12, 2015 12:52 PM, "erik quanstrom" <quanstro@quanstro.net> wrote:
> > so an interesting problem i run into from time to time is separately
> computing
> > the files added to and deleted from a directory in a shell script.  uniq
> doesn't
> > work for this.  certainly one can loop over two lists of files and do
> this easily,
> > but that seems dull and tedious.
> >
> > but as it turns out, one can compute the deleted and added files
> relatively
> > efficiently with diff in two steps.  obviously uniq -d gives us the
> union, so
> >
> >         fn ∩ {echo $$1 $$2 | sed 's/ /\n/g' | sort | uniq -d}
> 
> Isn't this called (set) intersection?
> 

; grep `{unicode ∩} /lib/unicode
002229	intersection

> > and uniq -u gives us not union
> >
> >         fn not∩ {echo $$1 $$2 | sed 's/ /\n/g' | sort | uniq -u}
> 
> this is (set) symmetric difference.

that sounds like a reasonable name, but not one i remembered.

- erik

Re: [9fans] fun with rc

[Christian Neukirchen]

erik quanstrom <quanstro@quanstro.net> writes:

> so an interesting problem i run into from time to time is separately computing
> the files added to and deleted from a directory in a shell script.  uniq doesn't
> work for this.  certainly one can loop over two lists of files and do
> this easily,
> but that seems dull and tedious.

Isn't this a canonical use case for comm(1)?

--
Christian Neukirchen  <chneukirchen@gmail.com>  http://chneukirchen.org

Re: [9fans] fun with rc

[erik quanstrom]

On Thu Mar 12 11:10:17 PDT 2015, chneukirchen@gmail.com wrote:
> erik quanstrom <quanstro@quanstro.net> writes:
>
> > so an interesting problem i run into from time to time is separately computing
> > the files added to and deleted from a directory in a shell script.  uniq doesn't
> > work for this.  certainly one can loop over two lists of files and do
> > this easily,
> > but that seems dull and tedious.
>
> Isn't this a canonical use case for comm(1)?

yes.  i hadn't thought of that.  that's trivial.

- erik