openssh unix dev - Jul 2022 - scp3.sh regression test fails if scp not already installed

Hi.

On Sat, 23 Jul 2022 at 05:34, rapier <rapier at psc.edu> wrote:
[...]
> The following patch checks to see if it's in the path or if it's
only in
> the openssh-portable build directory. I have a strong suspicion that
> this test will fail in debuild or rpmbuild environments but it will fail
> in a way that the test is run.
Thanks for the report.  I am fine with doing this in principle but
there's a few problems with the specifics of this patch:
> +scp_path=$(which -a scp)
> +if [ -x "$scp_path" ]
"which" is not POSIX, its semantics vary wildly and not all of them
have a "-a".  This test is probably also going to do the wrong thing
in the case where there are two scps on the path.  There's a
"have_prog" function in test-exec.sh in portable for this specific
purpose.

Also, the $() subshell syntax is not in the earlier POSIX standards
and doesn't work on older shells.  Solaris is particularly annoying in
this regard: /bin/sh on Solaris <=10 doesn't understand $() while on
Solaris 11 it's ksh in disguise and does but will complain about
backticks (under at least some conditions).
> || [ $scp_path == "*openssh-portable*" ]; then
What's the purpose of this?  In case the build dir is in the build
shell's path but not the login shell's?  There's no guarantee the
build directory is going to be named in a way that matches.  Also,
pattern matching with "==" is not POSIX (we would use a
"case" if we
needed to do this, but I think there's a better way here).

Instead, I think what we would need to do is check for scp in the path
over the ssh connection in a similar way to how have_prog does it.
> +       echo "scp not found in path. Skipping scp3 test."
> +       echo "*****************"
> +       exit
test-exec.sh has a "skip" function to do this in a consistent (and
shorter) way.

After a bit of fiddling, this seems to work, which I'll commit shortly:

$SSH -F $OBJ/ssh_proxy somehost \
    'IFS=":"; for i in $PATH;do [ -x "$i/scp" ]
&& exit 0; done; exit 1'
if [ $? -eq 1 ]; then
       skip "No scp on remote path."
fi

-- 
Darren Tucker (dtucker at dtucker.net)
GPG key 11EAA6FA / A86E 3E07 5B19 5880 E860  37F4 9357 ECEF 11EA A6FA (new)
    Good judgement comes with experience. Unfortunately, the experience
usually comes from bad judgement.

On Sat, 23 Jul 2022, Darren Tucker wrote:
> Solaris 11 it's ksh in disguise and does but will complain about
> backticks (under at least some conditions).
The accent-gravis form of command substitution is n?n-portable in
at least one important regard: the handling of double quotes.

In general, you can have double quotes *either* outside *or* inside
the accent-gravis command substitution?

	echo "foo  `echo bar`  baz"

	echo foo `echo "$bar"` baz

? but NOT both:

	echo "foo  `echo "$bar"` baz" # IS AN ERROR

This is because some shells priorise the accent-gravis over
the double quote, some shells priorise the double quote higher.

In general, it?s unsafe to use. I do use it in the build script
for mksh, because that one has to be portable, but for most uses
it might be sensible to require the shell to be somewhat capable.

(mksh/Build.sh requires the shell to have functions; some very
old ash variants lacked them, but this is where I draw the line?
GNU autoconf is generated so they can expand inline all the time,
but for hand-written scripts, this isn?t tenable.)

bye,
//mirabilos (mksh developer)
-- 
Infrastrukturexperte ? tarent solutions GmbH
Am Dickobskreuz 10, D-53121 Bonn ? http://www.tarent.de/
Telephon +49 228 54881-393 ? Fax: +49 228 54881-235
HRB AG Bonn 5168 ? USt-ID (VAT): DE122264941
Gesch?ftsf?hrer: Dr. Stefan Barth, Kai Ebenrett, Boris Esser, Alexander Steeg

                        ****************************************************
/?\ The UTF-8 Ribbon
??? Campaign against      Mit dem tarent-Newsletter nichts mehr verpassen:
??? HTML eMail! Also,     https://www.tarent.de/newsletter
??? header encryption!
                        ****************************************************

Hey, sorry for the late reply. I was at a conference and then got food 
poisoning so I've been spending way too long catching up.

On 7/23/22 12:33 AM, Darren Tucker wrote:
> 
> Thanks for the report.  I am fine with doing this in principle but
> there's a few problems with the specifics of this patch:
That's cool. I'm not really very good at shell scripting (at all) and I 
just went with the first thing I found that worked in my context. I 
submitted the patch as an example rather than anything I expected to 
have included.
>> || [ $scp_path == "*openssh-portable*" ]; then
> 
> What's the purpose of this?  In case the build dir is in the build
> shell's path but not the login shell's? 
Yup. I knew it was fragile but I didn't know of a better way of doing it.
> After a bit of fiddling, this seems to work, which I'll commit shortly:
> 
> $SSH -F $OBJ/ssh_proxy somehost \
>      'IFS=":"; for i in $PATH;do [ -x "$i/scp" ]
&& exit 0; done; exit 1'
> if [ $? -eq 1 ]; then
>         skip "No scp on remote path."
> fi
Sweet! I'll get that into my code base as well. Thanks for looking at this.

Chris