On my quad-core MacBook, I've used "yes" in four different terminals as a way to quickly test a 100% CPU load.

Yes (so to speak), you can also do 'yes &' multiple times to create lots of background load (and then 'pkill yes' before anything melts...)

I know its only an example but why

   yes | rm -r <>

and not

    rm -rf <>

I believe "yes" has become less useful over the years, with flags like force and quiet solves this now.

I think rm -rf would also suppress errors in addition to confirmation messages, so presumably if you want a "soft" force...

No it doesn't.

  > rm -rf foo/bar
  rm: foo/bar: Permission denied

If foo/bar does not exist, it should be silent. Via http://pubs.opengroup.org/onlinepubs/007904975/utilities/rm.... on the -f option:

> [...] Do not write diagnostic messages or modify the exit status in the case of nonexistent operands. [...]

Yes, but that's the only error it suppresses, and that's because "not existing" isn't really an error for a tool that's trying to delete something. In my example foo/bar existed but I didn't have permissions to delete it, and as you'll see it printed out that error.

I sometimes do

   yes "" | cat -n

to generate a sequence number.

Try `seq 1 n`

euske may be using one of the operating systems (one such being mentioned on this very page) where there is no seq utility.

    JdeBP ~ $seq
    ksh: seq: not found
    JdeBP ~ $

Of course, on the one mentioned there is jot.

And on bash you can simply use {1..n}:

    $ echo {1..5}
    1 2 3 4 5

If you have Bash 4.0 or above, you can include a step size:

    $ echo {1..10..2}
    1 3 5 7 9

"yes d | mail" is an excellent way to get rid of failed cron job mails.

Why wouldn't it be `mail | yes d` if you're trying to answer "d" to all of the input requests of mail?

Because you want all the d's to be sent to mail...