Test suite failures in 5.0.2-test-1 _and_ 5.0.2-test-3 on automatic build (was: Re: 5.0.2-test-2 is available)

[Axel Beckert <abe@deuxchevaux.org>]

Hi,

On Mon, Dec 02, 2013 at 04:15:02PM +0100, Axel Beckert wrote:
> 5.0.2-test-1 btw. had one test-suite failure when building inside a
> jenkins automatic build:
> http://jenkins.grml.org/view/Debian/job/zsh-binaries/90/architecture=i386/console
>
> As soon as I have 5.0.2-test-3 building, I'll check if that still
> happens. Just thought, I should give a heads up as soon as possible
> (since I've already waited to long to do so).

This issue is still there with 5.0.2-test-3:

http://jenkins.grml.org/view/Debian/job/zsh-binaries/architecture=i386/91/console

The following test fails:

[…]
../../Test/A05execution.ztst: starting.
Unable to change MONITOR option
*** /tmp/zsh.ztst.err.16395	Mon Dec  2 16:15:29 2013
--- /tmp/zsh.ztst.terr.16395	Mon Dec  2 16:15:29 2013
***************
*** 0 ****
--- 1 ----
+ (eval):setopt:1: can't change option: MONITOR
Test ../../Test/A05execution.ztst failed: error output differs from expected as shown above for:
  setopt MONITOR
  [[ -o MONITOR ]] || print -u $ZTST_fd 'Unable to change MONITOR option'
  repeat 2048; do (return 2 |
                   return 1 |
                   while true; do
                             false
                             break
                           done;
                   print "${pipestatus[@]}")
		   ZTST_hashmark
  done | sort | uniq -c | sed 's/^ *//'
Was testing: Check whether `$pipestatus[]' behaves.
../../Test/A05execution.ztst: test failed.
The following may (or may not) help identifying the cause:
  This test checks for a bug in `$pipestatus[]' handling.  If it breaks then
  the bug is still there or it reappeared. See workers-29973 for details.
../../Test/A06assign.ztst: starting.
[…]
../../Test/Y03arguments.ztst: all tests successful.
**************************************
41 successful test scripts, 1 failure, 0 skipped
**************************************

That code was last touched recently by the following two commits:

commit c3114a7735c85b79771e08bd156470bde1a36950
Author: Barton E. Schaefer <schaefer@zsh.org>
Date:   Sat Oct 26 16:11:40 2013 -0700

    31906: fix race-condition interaction of $pipestatus with job control
    
    printjob() should not reference oldjobtab for job numbers unless
    it is being called from bin_fg(). printjob() also must not attempt
    to update pipestats when called from bin_fg(). acquire_pgrp()
    should not loop infintely if the shell is not interactive. Update
    the $pipestatus stress test so that it also exercises the
    oldjobtab repair.

commit 65e131c838be6a300b41e3fa8259c281714e4918
Author: Barton E. Schaefer <schaefer@zsh.org>
Date:   Wed Oct 23 22:45:02 2013 -0700

    31879 plus misc.:  improve $pipestatus handling and add a test for it

Since this test does not fail on a normal package build, I suspect
it's related to being connected to a terminal or not.

> > 2. Builtins that are marked to be loaded from modules (using "zmodload")
> > no longer cause an error if the module was already loaded.  This was
> > apparently causing a problem with the previous test version.
> 
> Thanks for that fix. (Not yet verified, but I'm optimistic. :-)

Confirmed, thanks!

		Kind regards, Axel
-- 
/~\  Plain Text Ribbon Campaign                   | Axel Beckert
\ /  Say No to HTML in E-Mail and News            | abe@deuxchevaux.org  (Mail)
 X   See http://www.asciiribbon.org/              | abe@noone.org (Mail+Jabber)
/ \  I love long mails: http://email.is-not-s.ms/ | http://noone.org/abe/ (Web)