[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openpkg-dev
Subject:    Re: OpenSSH PATH problem
From:       Bill Campbell <bill () celestial ! com>
Date:       2005-03-29 20:44:35
Message-ID: 20050329204435.GA48760 () alexis ! mi ! celestial ! com
[Download RAW message or body]

On Tue, Mar 29, 2005, Ralf S. Engelschall wrote:
>On Tue, Mar 29, 2005, Bill Campbell wrote:
>
>> I ran into a problem at a customer's yesterday when I found that
>> openssh no longer prepends the OpenPKG bin directory to the PATH
>> on the server.  Looking back through my archives, this appears to
>> have happened between Release 2.1 and Release 2.2.  This causes
>> things like ``rsync -e ssh xxx'' to fail on the remote system or
>> to find invalid configuration if there's a vendor supplied
>> version of the program on the remote end that conflicts with the
>> OpenPKG version of the program.
>>
>> The attached patch fixes this in what I think is a logical manner
>> for OpenPKG Release 2.2, 2.3, and CURRENT.
>
>Unfortunately, no, it doesn't. Well, to be precise, it "fixes" it
>on those platforms where OpenSSH supports the --with-default-path
>functionality only!
>
>That's the reason why we explicitly removed any %{l_prefix} stuff there
>(see http://cvs.openpkg.org/chngview?cn=19660) for cross-platform
>consistency reasons. Because it is totally confusing if one some
>platforms one magically has the %{l_prefix} in path by default set
>by sshd(8) and on other platforms it isn't. Hence we decided to
>intentionally not try to add %{l_prefix}/bin there at all and instead
>have to life with the fact that there has to be either (1) a global
>shell profile setting PATH correctly, or (2) a local shell profile
>setting PATH correctly or (3) the user calls the command with an
>absolute path.
>
>All this I personally don't like very much, but to be honest, I do not
>see a solution. Even hacking sshd(8) is not possible because the reason
>why --with-default-path isn't working on all platforms is partly because
>on those login(1) is used by sshd(8) (and this way PATH is reset by the
>system after sshd forked off its child), etc.

OK.  I thought there had to be a reaon.  How about adding an option to
openssh, say with_dumbpath, that's normally set to 'no', so the person
building openssh can make an appropriate choice without having to hack the
spec file locally as a special case.

Bill
--
INTERNET:   bill@Celestial.COM  Bill Campbell; Celestial Systems, Inc.
UUCP:               camco!bill  PO Box 820; 6641 E. Mercer Way
FAX:            (206) 232-9186  Mercer Island, WA 98040-0820; (206) 236-1676
URL: http://www.celestial.com/

``The fact is that the Constitution was intended to protect us from
the government, and we cannot expect the government to enforce it
willingly'' -- Dave E. Hoffmann, Reason Magazine March 2002
______________________________________________________________________
The OpenPKG Project                                    www.openpkg.org
Developer Communication List                   openpkg-dev@openpkg.org
[prev in list] [next in list] [prev in thread] [next in thread]