Re: (zsh-4.3.2) Y03 test failure under Slackware Linux 9.0.0

Re: (zsh-4.3.2) Y03 test failure under Slackware Linux 9.0.0

[Eric De Mund]

Peter,

Eric De Mund wrote:
> I'm experiencing a "make check" failure in test Y03 when building
> zsh-4.3.2 under Slackware Linux 9.0.0 on a i686-pc-linux-gnu host, as I
> experienced back in March of this year (Message-ID:
> <17433.46602.590991.508063@bear.he.net>).

Peter Stephenson wrote:
] The failure is this:
] 
] *** /tmp/zsh.ztst.out.9272	Mon Nov 27 00:13:12 2006
] --- /tmp/zsh.ztst.tout.9272	Mon Nov 27 00:13:12 2006
] ***************
] *** 2,8 ****
]   line: {tst -o +o }{}
]   line: {tst -o +o }{}
]   MESSAGE:{no arguments}
] ! line: {tst +o }{}
] ! line: {tst +o -o }{}
] ! line: {tst +o -o }{}
]   MESSAGE:{no arguments}
] --- 2,10 ----
]   line: {tst -o +o }{}
]   line: {tst -o +o }{}
]   MESSAGE:{no arguments}
] ! line: {tst -+}{}
] ! MESSAGE:{no arguments}
] ! line: {tst -+}{}
] ! MESSAGE:{no arguments}
] ! line: {tst -+}{}
]   MESSAGE:{no arguments}
] Test ./Y03arguments.ztst failed: output differs from expected as shown above for:
]  tst_arguments -+o
]  comptest $'tst -\t\t\t\C-w\C-w+\t\t\t'
] Was testing: option beginning with + and -.
] ./Y03arguments.ztst: test failed.
]
] [...]
]
] My best guess is that the two \C-w's in the test aren't wiping the -
] from the previous line, "tst -o +o ". That suggests $WORDCHARS doesn't
] include "-" or it isn't having an effect for some reason. That would
] mean the test actually started the second set of completions with "tst
] -+" on the line, which then behaved the way you'd expect, so the
] failure isn't in the completion part at all. This seems more likely
] than the "-+" popping out of a failed completion.
]
] What happens if you start from "zsh -f" (and "bindkey -e" if neces-
] sary) and type:
]
] % tst -o +o\C-w\C-w
]
] does the - disappear (as it should)?  If not, what does "print -r
] $WORDCHARS" show?

On the Slackware 9.0.0 host in question:

    bear:~% cat /etc/slackware-version
    Slackware 9.0.0
    bear:~% ~/bin/config.guess
    i686-pc-linux-gnu
    bear:~% 

, the - (minus) does not disappear; I'm left with:

    bear:~% tst -

However, WORDCHARS consists solely of an _ (underscore):

    bear:~% print -r $WORDCHARS
    -
    bear:~% 

When I set WORDCHARS to that which it's set to on another host, my
cygwin host:

    cowberry:~% uname -s
    CYGWIN_NT-5.1
    cowberry:~% print -r $WORDCHARS
    *?_-.[]~=/&;!#$%^(){}<>

, then your requested test only leaves me with "tst ":

    bear:~% WORDCHARS='*?_-.[]~=/&;!#$%^(){}<>'
    bear:~% tst 
                ^---- cursor is here
    bear:~% print -r $WORDCHARS
    *?_-.[]~=/&;!#$%^(){}<>
    bear:~% 

So do I understand correctly that WORDCHARS should be set to more than _
(a lone underscore) in a "zsh -f" shell?

Let me know if you need me to perform any other tests.

[several minutes later]

Ok, I believe I've solved this. My build environment for zsh-4.3.2 was a
zsh-4.2.5 shell with my ~/.{zshenv,zshrc,zlogin} files invoked, which
includes this variable setting: WORDCHARS="_".

When I removed my WORDCHARS setting (and logged out and logged back in),
WORDCHARS reverted to what I presume is the default:

    bear:~% print -r $WORDCHARS
    *?_-.[]~=/&;!#$%^(){}<>
    bear:~% 

I then unpacked zsh-4.3.2.tar.bz into a brand new directory, configured
it, and performed a "make" and a "make check". All the tests in
zsh-4.3.2's "make check" passed.

It doesn't seem right that a "make" or "make check" of a new zsh, if
done from within the environment of an old zsh, should require a
"zsh -f" environment. However, that's your call. It seems to me that if
the Y03 test (or any other test) requires the environment to be set a
certain way, it should set it up itself. My sense here is that the
$WORDCHARS value is presumed set appropriately rather than being set
explicitly.

Thank you for helping me drill down on this problem; I appreciate it
very much.

Regards,
Eric
--
Eric De Mund              |      Ixian Systems      | 
email: <ead@ixian.com>    | 650 Castro St, #120-210 | ICQ: 811788
http://www.ixian.com/ead/ | Mountain View, CA 94041 | Y!M: ead0002

Re: (zsh-4.3.2) Y03 test failure under Slackware Linux 9.0.0

[Peter Stephenson]

Eric De Mund wrote:
> Ok, I believe I've solved this. My build environment for zsh-4.3.2 was a
> zsh-4.2.5 shell with my ~/.{zshenv,zshrc,zlogin} files invoked, which
> includes this variable setting: WORDCHARS="_".

Yes, if I set that and export WORDCHARS I get exactly your result.  Do
you have WORDCHARS exported (it usually isn't)?  This should fix it
if so.

Thanks for looking.

Index: Test/ztst.zsh
===================================================================
RCS file: /cvsroot/zsh/zsh/Test/ztst.zsh,v
retrieving revision 1.25
diff -u -r1.25 ztst.zsh
--- Test/ztst.zsh	23 Sep 2006 06:55:29 -0000	1.25
+++ Test/ztst.zsh	28 Nov 2006 22:06:05 -0000
@@ -30,6 +30,9 @@
 [[ -n $LC_MESSAGES ]] && LC_MESSAGES=C
 [[ -n $LANG ]] && LANG=C
 
+# Don't propagate variables that are set by default in the shell.
+typeset +x WORDCHARS
+
 # Set the module load path to correspond to this build of zsh.
 # This Modules directory should have been created by "make check".
 [[ -d Modules/zsh ]] && module_path=( $PWD/Modules )

-- 
Peter Stephenson <p.w.stephenson@ntlworld.com>
Web page now at http://homepage.ntlworld.com/p.w.stephenson/

(zsh-4.3.2) Y03 test failure under Slackware Linux 9.0.0

[Eric De Mund]

People,

I'm experiencing a "make check" failure in test Y03 when building
zsh-4.3.2 under Slackware Linux 9.0.0 on a i686-pc-linux-gnu host, as I
experienced back in March of this year (Message-ID:
<17433.46602.590991.508063@bear.he.net>).

The build seemed to succeed as did the install, so I'm able to go
forward with my use of zsh-4.3.2. However, perhaps a zsh worker may wish
to look into this.

If so, my Y03 test failure can be browsed at
<http://www.ixian.com/ead/tmp/zsh-4.3.2/make-check.out>. As can every
aspect of this build be browsed at
<http://www.ixian.com/ead/tmp/zsh-4.3.2/>; I've made all the directories
and files world-readable.

FYI, the commands I issued were:

    % bzcat ~/tmp/zsh-4.3.2.tar.bz2 | tar -xvf -
        [output elided]
    % cd zsh-4.3.2
    % ./configure --prefix=$HOME/local --disable-zprofile LDFLAGS="-s"
        [output elided]
    % make |& tee make.out
        [output elided]
    % make check |& tee make-check.out
        [output elided]
    % make -k check |& tee make-check-2.out
        [output elided]
    % make install |& tee make-install.out
        [output elided]

Thank you for a superb shell; I've been a happy user of it since 1989.

Regards,
Eric
--
Eric De Mund
email: <ead@ixian.com>

Re: (zsh-4.3.2) Y03 test failure under Slackware Linux 9.0.0

[Peter Stephenson]

Eric De Mund wrote:
> I'm experiencing a "make check" failure in test Y03 when building
> zsh-4.3.2 under Slackware Linux 9.0.0 on a i686-pc-linux-gnu host, as I
> experienced back in March of this year (Message-ID:
> <17433.46602.590991.508063@bear.he.net>).

The failure is this:

*** /tmp/zsh.ztst.out.9272	Mon Nov 27 00:13:12 2006
--- /tmp/zsh.ztst.tout.9272	Mon Nov 27 00:13:12 2006
***************
*** 2,8 ****
  line: {tst -o +o }{}
  line: {tst -o +o }{}
  MESSAGE:{no arguments}
! line: {tst +o }{}
! line: {tst +o -o }{}
! line: {tst +o -o }{}
  MESSAGE:{no arguments}
--- 2,10 ----
  line: {tst -o +o }{}
  line: {tst -o +o }{}
  MESSAGE:{no arguments}
! line: {tst -+}{}
! MESSAGE:{no arguments}
! line: {tst -+}{}
! MESSAGE:{no arguments}
! line: {tst -+}{}
  MESSAGE:{no arguments}
Test ./Y03arguments.ztst failed: output differs from expected as shown above for:
 tst_arguments -+o
 comptest $'tst -\t\t\t\C-w\C-w+\t\t\t'
Was testing: option beginning with + and -.
./Y03arguments.ztst: test failed.

I can't this to happen with 4.3.2 on my Fedora machine, or with the
latest shell there or on Solaris.  It would be interesting to know if
it's still present in the version in the archive at the moment.

My best guess is that the two \C-w's in the test aren't wiping the -
from the previous line, "tst -o +o ".  That suggests $WORDCHARS doesn't
include "-" or it isn't having an effect for some reason.  That would
mean the test actually started the second set of completions with "tst
-+" on the line, which then behaved the way you'd expect, so the failure
isn't in the completion part at all.  This seems more likely than the
"-+" popping out of a failed completion.

What happens if you start from "zsh -f" (and "bindkey -e" if necessary)
and type:

% tst -o +o\C-w\C-w

does the - disappear (as it should)?  If not, what does "print -r
$WORDCHARS" show?

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070


To access the latest news from CSR copy this link into a web browser:  http://www.csr.com/email_sig.php