NON-SOURCE POSTINGS CONSIDERED HARMFUL!

[Dan Bernstein]

Well, ain't this cute. Tom posts 28 lines of Perl to do FOO. I post 7
lines of sh+tools to do essentially the same thing, and I mention that
it doesn't do trivial feature BAR. Tom says ``Your script doesn't at all
do the same thing''; when pressed, he admits that he's only talking
about BAR.

Fine, here's a simple sh script to do FOO + BAR (modulo the minor
caveats from before). It's essentially the same as the previous script,
with a ``which'' loop in the style people seem to like. Sheesh.

Face it, Tom, it takes twenty words of sh code (including | tokens) to
combine six *appropriate* tools to list repeated names in $PATH. It
takes much, much more than that to solve the problem in Perl without
outside help. The only rational conclusion is that Perl is the wrong
tool for the job.

#!/bin/sh
dirs=`echo "$PATH" | tr : '\012'`

# List all files in path. Look back to front. Sort, removing duplicates.
# Remove all ls info except filename, and select executables. Find reps.
for i in `ls -ilFL $dirs | rev | sort -u \
| sed -n -e 's/ .*//' -e 's/^\*//p' | uniq -d | rev`; do

# Print the repeats, in some stilted format that tchrist at convex.com wants.
# Yes, this is unnecessary duplication of a function already available in
# other programs, and it reduces the usefulness of this interface. Oh, well.
  echo -n "$i: "
  for j in $dirs; do
    test -f "$j/$i" && echo -n "$j/$i "
  done
  echo ''
done

> You assume you can tell me what I wrote my script to do, and then you go
> on to redefine this.  For this they invented the word arrogant, and then
> tacked ignorant on to the end of it.

Oh, fgs. You described your script as doing FOO, and I described mine as
doing FOO. At every stage in this discussion I have been very careful to
point out whose script does what, to note the differences (like BAR),
etc. I never tried to redefine what yours did, but you said that mine
didn't ``at all do the same thing'' as yours---an obviously false
statement, intended only to put me down because you know you've lost on
technical merits. Take a step back, Tom: who's being arrogant here?

> The problem was to list the full path names of all names in 
> collision.

I don't see ``full path names'' in the original problem statement.

> Nothing less than this solves the problem.  

Sure, Tom, whatever you say.

---Dan