Re: OpenSSH 4.7: call for testing.

On Fri, Aug 17, 2007 at 12:10:06PM +0200, Alan Barrett wrote:
> On Thu, 16 Aug 2007, Darren Tucker wrote:
> > Running the regression tests supplied with Portable does not require
> > installation and is a simply:
> >
> > $ ./configure && make tests
>
> I tested on NetBSD-current/i386 (version 4.99.27).
>
> There was no configure script in the code that I obtained from
> "cvs -d anoncvs@anoncvs.mindrot.org:/cvs checkout openssh".
>
> I managed to generate a configure script via "autoreconf",
> and the configure script appeared to work.

That's expected, and covered in the INSTALL file (which is a little
misnamed given that it covers both building and installing):

Autoconf:

If you modify configure.ac or configure doesn't exist (eg if you checked
the code out of CVS yourself) then you will need autoconf-2.61 to rebuild
the automatically generated files by running "autoreconf". Earlier
version may also work but this is not guaranteed.

http://www.gnu.org/software/autoconf/

(and yeah, I noticed the missing "s" on "version")

> make tests failed with the following error:
>
> run test connect.sh ...
> Missing privilege separation directory: /var/empty
> FATAL: sshd_proxy broken
> *** Error code 1
>
> I created /var/empty (as root) and then ran "make tests" again (as an
> unprivileged user).

sshd only checks that directory if use_privsep is enabled, but it
still does so when sshd is running as a regular user. Perhaps it
should unset use_privsep in that case.

> This time, "make tests" complained about some unrecognised syntax in my
> .ssh/config file. OK, the file did contain syntax that I didn't expect
> an unpatched vesion of openssh to understand, but it seems liek an error
> for the tests to use whatever random contant I happen to have in my
> configuration file. Surely the tests should use "-F /dev/null", or "-F
> ${special_configuration_file}"?

For the most part, it does use its own config files which it creates in
a working directory during the test and removes afterward (see, eg,
regress/connect.sh).

It appears that just the multiplex tests in particular don't, but that's
easily fixable, although that won't be until after the release.

> I moved by .ssh directory aside and tried "make tests" again.
>
> This time, "make tests" succeeded. However, it created an empty
> $HOME/.ssh directory. I think it's rude for a test suite to create any
> non-temporary files or directories.

That's a reasonable point, however offhand I can't think of a way to stop
it.

Thanks for the testing and report.

--
Darren Tucker (dtucker at zip.com.au)
GPG key 8FF4FA69 / D9A3 86E9 7EEE AF4B B2D4 37C9 C982 80C7 8FF4 FA69
Good judgement comes with experience. Unfortunately, the experience
usually comes from bad judgement.
_______________________________________________
openssh-unix-dev mailing list
openssh-unix-dev@mindrot.org
https://lists.mindrot.org/mailman/li...enssh-unix-dev