On 22 Feb 2000, Chris G. Demetriou wrote:

: > : what's the value in actually ditching it?
: > 
: > like, rsh has no way to actually *fail immediately* if the connection is
: > refused because rshd ain't running?
: 
: Uh, well, so, that's a problem in the historical rsh protocol and/or
: TCP.  (no way to differentiate between unable to connect, and
: temporarily unable to connect because of resource limitation issues)

Well, you asked what the value was--there is actual value in at least
providing a way to turn it off.

: However, if you want to interoperate with other implementations in the
: way that people come to expect, then you don't necessarily want to tweak
: historic behaviour.  I think my personal feeling would be that if you're
: using anything involving rsh, anyway, you've already got lossage.

My personal beef is with rsh from the command line, which functions the same
way (and, I believe, does it via a call to rcmd()).  Now, were this made
some kind of one-off environment variable only for rsh(1) or somesuch, I
would have no problem with it.

-- 
-- Todd Vierling (tv@pobox.com)