svn commit: r279603 - in head: bin/rcp usr.bin/rlogin usr.bin/rsh

Harald Schmalzbauer h.schmalzbauer at omnilan.de
Fri Mar 6 07:29:50 UTC 2015


 Bezüglich Garrett Cooper's Nachricht vom 05.03.2015 19:47 (localtime):
> On Mar 5, 2015, at 10:21, Hans Ottevanger <hans at beastielabs.net> wrote:
>
>> On 03/05/15 13:21, Slawa Olhovchenkov wrote:
>>> On Thu, Mar 05, 2015 at 02:48:29PM +0300, Gleb Smirnoff wrote:
>>>
>>>> On Wed, Mar 04, 2015 at 10:01:45PM +0000, Baptiste Daroussin wrote:
>>>> B> Author: bapt
>>>> B> Date: Wed Mar  4 22:01:44 2015
>>>> B> New Revision: 279603
>>>> B> URL: https://svnweb.freebsd.org/changeset/base/279603
>>>> B>
>>>> B> Log:
>>>> B>   r* commands are not precious anymore
>>>> B>
>>>> B> Modified:
>>>> B>   head/bin/rcp/Makefile
>>>> B>   head/usr.bin/rlogin/Makefile
>>>>
>>>> I guess when they are going to be not precious enough to be removed? :)
>>>>
>>>> In modern world of ssh and https, does any OS require them in base?
>>> yes.
>>> Some telecom equipment require rlogin.
>>> _______________________________________________
>> Considering that the r-commands are not particularly large and also not really a maintenance nightmare, a would just keep them. They are (still) more or less part of the standard Unix toolbox, as perceived by end-users, and you had better not make life too difficult for them. The same is true for telnet.
>>
>> I see these tools in use regularly, e.g. to control measurement equipment programmatically. Due to the price tag of those instruments, that won't change overnight. The usage is limited to a LAN however, nobody I know uses these tools over the public Internet anymore.
>>
>> As far as I know only OpenBSD got rid of these tools up to now. Most other Unix(-like) systems still have them.
>>
>> And if they absolutely have to go, what happens to the corresponding daemons in /usr/libexec (rshd and rlogind)?
> Why not just move them to ports so the people that need them can have them…?

Installing ports on non-internet-connected machines, where you have your
self-brew pakcgae carried to the destination machine via scp or disk or
UFD, is a nightmare. We do not have pkg(8) in base. pkg(8) has circular
dependency on itself and every once in a while I have hassels with my
build hosts because ports infrastructre inists on a different pkg(8)
version :-(
Before pkg(8) is in base, moving any thing from base to ports is no
option IMHO.

I use rsh(1) for zfs management over LAN (most times dedicated SAN
VLAN). rsh(1) makes great zfs-snapshot-based backup-push methods easily
deployable! rlogin/telnet is nice to have for novices, reading ancient
Unix/FreeBSD/nameit books and getting in first contact, IMHO.

I need, rsh(1) please don't touch, we have "WITHOUT_RCMDS" for
src.conf(5), which I also use, but for completely different setups.

Regarding telnet(1) vs. nc(1): I can't second that nc(1) isn't _the_
appropriate tool for network connection testings. telnet(1) is often
absued for such tests, but that's not what it's intended for, nc(1) does
the job! Nevertheless, I vote for keeping telnet(1) in base.

Thanks,

-Harry

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 196 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freebsd.org/pipermail/svn-src-head/attachments/20150306/8c95997c/attachment.sig>


More information about the svn-src-head mailing list