Yo All! Well I work on a diverse number of OS's with a diverse number of clients. Some use F-Secure, SecureCRT, PuTTY, SSH.COM. OpenSSH, etc. with a wide variety of versions between each, some from source, some from rpms, etc... Basically a lot of legacy stuff that no one has the time to update. In fact I am working on a couple of OpenSSH config problems in the last few days. Sometimes we want to use SSH2 for the better antispoofing and sometimes SSH1 for scp compatibility and sometimes we have to change between versions to use RSA and DSA keys and do it from scripts. Having -1 and -2 would make life a lot easier, if only for testing purposes. Even a "-o [option]" with scp would be nice. A lot of these different products only interoperate in certain modes so I have to try a lot of things. I would also like to argue for more orthogonality between sch, ssh and sshd. Why do you turn on verbosity and debugging with "-v" in scp and ssh, but use "-d" with sshd? Why do you select a port with scp and sshd with "-P port" but with "-p port" on sshd? My poor little brain has enough to remember without all these changes that make little difference? And having to muck with config files leaves a lot of mess behind. RGDS GARY --------------------------------------------------------------------------- Gary E. Miller Rellim 20340 Empire Ave, Suite E-3, Bend, OR 97701 gem at rellim.com Tel:+1(541)382-8588 Fax: +1(541)382-8676
At 3:46 PM -0700 9/8/00, Gary E. Miller wrote:>I would also like to argue for more orthogonality between scp, >ssh and sshd.I suspect you did not want to ask for "more orthogonality"...>Why do you turn on verbosity and debugging with "-v" in scp >and ssh, but use "-d" with sshd? Why do you select a port with >scp and sshd with "-P port" but with "-p port" on sshd?some of this is because of what the different commands do. scp is patterned after cp, and thus it pretty much has to have -r as "recursive" and -p as "preserve". Thus, you would really irritate people if you decided to use '-p' to select a port number. It would be more consistent to use -P for port, because cp does not use -P for anything. sshd is probably mimicing assorted other system daemons, which are much more likely to use -p to select a port instead of -P. (that's off the top of my head, I haven't actually checked all possible system daemons to see if that is strictly true). Thus, it probably IS more consistent for sshd to use -p. ssh itself is trying to look like rsh. So, it is more important for it's options to match rsh's options than for them to match scp's options. --- Garance Alistair Drosehn = gad at eclipse.acs.rpi.edu Senior Systems Programmer or drosih at rpi.edu Rensselaer Polytechnic Institute
Yo Garance! On Fri, 8 Sep 2000, Garance A Drosihn wrote:> At 3:46 PM -0700 9/8/00, Gary E. Miller wrote: > >I would also like to argue for more orthogonality between scp, > >ssh and sshd. > > I suspect you did not want to ask for "more orthogonality"...Yes I did. This is the lingo CPU designers us when they mean that they want a class of instructions to look similar.> >Why do you turn on verbosity and debugging with "-v" in scp > >and ssh, but use "-d" with sshd? Why do you select a port with > >scp and sshd with "-P port" but with "-p port" on sshd? > > some of this is because of what the different commands do.But we can do better than the old ways! Well how about for all of scp, ssh and sshd: -1 conflicts with nothing and looks like SSH.com -2 conflicts with nothing and looks like SSH.com -d conflicts with nothing and turns on debug stuff -P [port] to set the port to use -o [options] to set other options All the other stuff could stay for back compatibility. RGDS GARY --------------------------------------------------------------------------- Gary E. Miller Rellim 20340 Empire Ave, Suite E-3, Bend, OR 97701 gem at rellim.com Tel:+1(541)382-8588 Fax: +1(541)382-8676
On Sat, 9 Sep 2000, Gert Doering wrote:> Hi, > > On Fri, Sep 08, 2000 at 06:14:01PM -0700, Gary E. Miller wrote: > > On Fri, 8 Sep 2000, Garance A Drosihn wrote: > > > At 3:46 PM -0700 9/8/00, Gary E. Miller wrote: > > > >I would also like to argue for more orthogonality between scp, > > > >ssh and sshd. > > > > > > I suspect you did not want to ask for "more orthogonality"... > > Yes I did. This is the lingo CPU designers us when they mean that > > they want a class of instructions to look similar. > > Actually, "orthogonal" in mathemathics and physics means exactly the > opposite: a set of vectors being orthogonal means all of them are at > right angles to each others, the scalar product being zero... > > gertYes, but in computer sciences "orthogonal" means that all the software components are independent of each other and that they behave the same way in any circumstance (which was what the originating poster meant). -- Pekka Savola "Tell me of difficulties surmounted, Pekka.Savola at netcore.fi not those you stumble over and fall"
On Fri, Sep 08, 2000 at 03:46:57PM -0700, Gary E. Miller wrote:> Having -1 and -2 would make life a lot easier, if only for testing > purposes. Even a "-o [option]" with scp would be nice. A lot of > these different products only interoperate in certain modes so I > have to try a lot of things.if you switch between protocols all the time, then editing .ssh/config is probably the 'right thing'. if you want to switch occasionally, then it's easy to type -o protocol=2> I would also like to argue for more orthogonality between sch, ssh and > sshd. Why do you turn on verbosity and debugging with "-v" in scp > and ssh, but use "-d" with sshd? Why do you select a port with > scp and sshd with "-P port" but with "-p port" on sshd?history and compatibility with rcp. but there are much more imporant things todo with openssh than adding millions of cmdline options. imho, scp needs only -o and the options from rcp. everything else can be removed. -markus
Hi, On Fri, Sep 08, 2000 at 06:14:01PM -0700, Gary E. Miller wrote:> On Fri, 8 Sep 2000, Garance A Drosihn wrote: > > At 3:46 PM -0700 9/8/00, Gary E. Miller wrote: > > >I would also like to argue for more orthogonality between scp, > > >ssh and sshd. > > > > I suspect you did not want to ask for "more orthogonality"... > Yes I did. This is the lingo CPU designers us when they mean that > they want a class of instructions to look similar.Actually, "orthogonal" in mathemathics and physics means exactly the opposite: a set of vectors being orthogonal means all of them are at right angles to each others, the scalar product being zero... gert -- Gert Doering
At 12:56 PM +0200 9/9/00, Gert Doering wrote:>Hi, > >On Fri, Sep 08, 2000 at 06:14:01PM -0700, Gary E. Miller wrote: > > On Fri, 8 Sep 2000, Garance A Drosihn wrote: > > > At 3:46 PM -0700 9/8/00, Gary E. Miller wrote: > > > >I would also like to argue for more orthogonality between scp, > > > >ssh and sshd. > > > > > > I suspect you did not want to ask for "more orthogonality"... > > > > Yes I did. This is the lingo CPU designers us when they mean that > > they want a class of instructions to look similar. > >Actually, "orthogonal" in mathemathics and physics means exactly the >opposite: a set of vectors being orthogonal means all of them are at >right angles to each others, the scalar product being zero...And indeed, I come from a physics and math background... :-) (many years ago). I still tend to use it the same way in computer science contexts, although that might not be standard practice. If I were to say the options in a set of commands should be orthogonal, I would mean that the options for one command would be unrelated to the options for another command in that set. If I want the options to be the same across a set of commands, I would much rather use the word "consistent", simply because everyone will have the same idea of what that means. We now return you to the original topic, instead of this orthogonal digression... :-) --- Garance Alistair Drosehn = gad at eclipse.acs.rpi.edu Senior Systems Programmer or drosih at rpi.edu Rensselaer Polytechnic Institute