On Thu, 12 Jul 2001, Greywolf wrote:

> Date: Thu, 12 Jul 2001 18:28:04 -0700 (PDT)
> From: Greywolf <greywolf@starwolf.com>

> # p.s. Tnx for your suggestion about "cmp".  But having both "diff" and "cmp"
> # is sort of like having both "ls" and (a currently non-existent) "dir", yes?
>
> Not really.
>
> Diff will actually show the differences; cmp will merely abort at the first
> one and bitch mightily.  Sometimes it is only wanted to know whether or not

cmp -l <file1> <file>

will dump every difference it finds; that's what I used.


> ...and why it reads BOTH files ALL THE WAY into memory is beyond me.  You'd
> think they could do it chunks at a time (like any sane utility).  512K/file
> strikes me as plenty of data to operate on, maybe 1M/file.

Even if memory constrained, I wonder if they ever heard of "lseek(2)" ?

Well, I'd like not to just complain, but to actually fix the problem.  But
since it's GNU, looks like I'll need to take it up with the GNU people, or
write a BSD-diff.

-Mike