Re: zsh 4.3.9-dev-2

Re: zsh 4.3.9-dev-2

[Bart Schaefer]

On Apr 17,  6:04pm, Peter Stephenson wrote:
} Subject: zsh 4.3.9-dev-2
}
} ftp://ftp.zsh.org/pub/development/zsh-4.3.9-dev-2.tar.gz
} ftp://ftp.zsh.org/pub/development/zsh-4.3.9-dev-2-doc.tar.gz
} 
} Please try it and let zsh-workers know if there are problem.

Current CVS checkout builds cleanly and passes all tests starting from
"make distclean" on Mac OSX (Darwin 9.6.0).

Trying the same thing on Linux 2.6.9-78.0.1.plus.c4spm (CentOS 4) I get

echo 'timestamp for *.mdd files' > ../Src/modules.stamp
make[3]: *** No rule to make target `headers'
make[2]: *** [headers] Error 1
make[1]: *** [headers] Error 2

One significant difference is that on Linux I'm using a separate build
tree rather than compiling directly in the source tree.

I tried starting over from scratch and got a different problem:

torch% ../source/configure
(this all goes fine)
torch% make
cd .. && /bin/sh $top_srcdir/Src/mkmakemod.sh Src Makemod
creating Src/Makemod.in
config.status: creating Src/Makemod
creating Src/Builtins/Makefile.in
config.status: creating Src/Builtins/Makefile
creating Src/Modules/Makefile.in
config.status: creating Src/Modules/Makefile
creating Src/Zle/Makefile.in
config.status: creating Src/Zle/Makefile
(etc. until eventually)
gcc -c -I.  -DHAVE_CONFIG_H -Wall -Wmissing-prototypes -O2  -o signames.o signames.c
In file included from signames.c:6:
zsh.mdh:34:25: builtin.epro: No such file or directory
zsh.mdh:35:24: compat.epro: No such file or directory
zsh.mdh:36:22: cond.epro: No such file or directory
zsh.mdh:37:22: exec.epro: No such file or directory
zsh.mdh:38:22: glob.epro: No such file or directory
zsh.mdh:39:27: hashtable.epro: No such file or directory
zsh.mdh:40:22: hist.epro: No such file or directory
zsh.mdh:41:22: init.epro: No such file or directory
zsh.mdh:42:23: input.epro: No such file or directory
zsh.mdh:43:22: jobs.epro: No such file or directory
zsh.mdh:44:21: lex.epro: No such file or directory
zsh.mdh:45:26: linklist.epro: No such file or directory
zsh.mdh:46:22: loop.epro: No such file or directory
zsh.mdh:47:22: math.epro: No such file or directory
zsh.mdh:48:21: mem.epro: No such file or directory
zsh.mdh:49:24: module.epro: No such file or directory
zsh.mdh:50:25: options.epro: No such file or directory
zsh.mdh:51:24: params.epro: No such file or directory
zsh.mdh:52:23: parse.epro: No such file or directory
zsh.mdh:53:25: pattern.epro: No such file or directory
zsh.mdh:54:24: prompt.epro: No such file or directory
zsh.mdh:55:25: signals.epro: No such file or directory
zsh.mdh:57:22: sort.epro: No such file or directory
zsh.mdh:58:24: string.epro: No such file or directory
zsh.mdh:59:23: subst.epro: No such file or directory
zsh.mdh:60:22: text.epro: No such file or directory
zsh.mdh:61:23: utils.epro: No such file or directory
zsh.mdh:62:23: watch.epro: No such file or directory
make[2]: *** [signames.o] Error 1
make[2]: Target `modobjs' not remade because of errors.
make[1]: *** [modobjs] Error 2

and then a bit later (because I'm using "make -k")

gcc -c -I.  -DHAVE_CONFIG_H -DMODULE -Wall -Wmissing-prototypes -O2 -fPIC -o errnames..o errnames.c
In file included from system.mdh:17,
                 from errnames.c:6:
zsh.mdh:34:25: builtin.epro: No such file or directory
zsh.mdh:35:24: compat.epro: No such file or directory
zsh.mdh:36:22: cond.epro: No such file or directory
zsh.mdh:37:22: exec.epro: No such file or directory
zsh.mdh:38:22: glob.epro: No such file or directory
zsh.mdh:39:27: hashtable.epro: No such file or directory
zsh.mdh:40:22: hist.epro: No such file or directory
zsh.mdh:41:22: init.epro: No such file or directory
zsh.mdh:42:23: input.epro: No such file or directory
zsh.mdh:43:22: jobs.epro: No such file or directory
zsh.mdh:44:21: lex.epro: No such file or directory
zsh.mdh:45:26: linklist.epro: No such file or directory
zsh.mdh:46:22: loop.epro: No such file or directory
zsh.mdh:47:22: math.epro: No such file or directory
zsh.mdh:48:21: mem.epro: No such file or directory
zsh.mdh:49:24: module.epro: No such file or directory
zsh.mdh:50:25: options.epro: No such file or directory
zsh.mdh:51:24: params.epro: No such file or directory
zsh.mdh:52:23: parse.epro: No such file or directory
zsh.mdh:53:25: pattern.epro: No such file or directory
zsh.mdh:54:24: prompt.epro: No such file or directory
zsh.mdh:55:25: signals.epro: No such file or directory
zsh.mdh:57:22: sort.epro: No such file or directory
zsh.mdh:58:24: string.epro: No such file or directory
zsh.mdh:59:23: subst.epro: No such file or directory
zsh.mdh:60:22: text.epro: No such file or directory
zsh.mdh:61:23: utils.epro: No such file or directory
zsh.mdh:62:23: watch.epro: No such file or directory
In file included from errnames.c:6:
system.mdh:31:24: system.epro: No such file or directory
make[3]: *** [errnames..o] Error 1
make[3]: Target `modules' not remade because of errors.
make[2]: *** [modules] Error 1
make[1]: *** [modules] Error 2
make[1]: Target `all' not remade because of errors.
make: *** [all] Error 1

-- 

Re: zsh 4.3.9-dev-2

[Bart Schaefer]

On Apr 17,  2:22pm, Bart Schaefer wrote:
}
} One significant difference is that on Linux I'm using a separate build
} tree rather than compiling directly in the source tree.
} 
} I tried starting over from scratch and got a different problem:

I just tried from scratch again a third time and this time it worked.
Must be some sort of timestamping problem on files checked out of CVS.

Re: zsh 4.3.9-dev-2

[Peter Stephenson]

On Fri, 17 Apr 2009 14:36:18 -0700
Bart Schaefer <schaefer@brasslantern.com> wrote:
> On Apr 17,  2:22pm, Bart Schaefer wrote:
> }
> } One significant difference is that on Linux I'm using a separate build
> } tree rather than compiling directly in the source tree.
> } 
> } I tried starting over from scratch and got a different problem:
> 
> I just tried from scratch again a third time and this time it worked.
> Must be some sort of timestamping problem on files checked out of CVS.

If it looks like it doesn't affect the tarball, I'll ignore it.

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

Re: zsh 4.3.9-dev-2, two test failures on NetBSD

[Paul Ackersviller]

I actually had three tests fail from the tarball, but one mysteriously
went away after using the cvs sources.  These are likely to be
longstanding issues, as I don't recall having testing here before.

./A01grammar.ztst: starting.
*** /tmp/zsh.ztst.err.2701	Mon Apr 20 01:48:54 2009
--- /tmp/zsh.ztst.terr.2701	Mon Apr 20 01:48:54 2009
***************
*** 0 ****
--- 1 ----
+ stty: TIOCGETD: Inappropriate ioctl for device
Test ./A01grammar.ztst failed: error output differs from expected as shown above for:
  (exec -l /bin/sh -c 'echo $0')
Was testing: `exec' with -l option
./A01grammar.ztst: test failed.

./V01zmodload.ztst: starting.
*** /tmp/zsh.ztst.err.15666	Mon Apr 20 01:50:13 2009
--- /tmp/zsh.ztst.terr.15666	Mon Apr 20 01:50:13 2009
***************
*** 0 ****
--- 1,2 ----
+ (eval):3: module `zsh/terminfo' has no such feature: `b:echoti': autoload cancelled
+ (eval):3: module `zsh/terminfo' has no such feature: `p:terminfo': autoload cancelled
Test ./V01zmodload.ztst failed: error output differs from expected as shown above for:
 for m in $mods
 do
   zmodload -i $m || mods[(r)$m]=()
 done
Was testing: Test loading of all compiled modules
./V01zmodload.ztst: test failed.

Re: zsh 4.3.9-dev-2, two test failures on NetBSD

[Peter Stephenson]

On Mon, 20 Apr 2009 04:00:18 +0000
Paul Ackersviller <pda@sdf.lonestar.org> wrote:
> ./A01grammar.ztst: starting.
> *** /tmp/zsh.ztst.err.2701	Mon Apr 20 01:48:54 2009
> --- /tmp/zsh.ztst.terr.2701	Mon Apr 20 01:48:54 2009
> ***************
> *** 0 ****
> --- 1 ----
> + stty: TIOCGETD: Inappropriate ioctl for device
> Test ./A01grammar.ztst failed: error output differs from expected as shown above for:
>   (exec -l /bin/sh -c 'echo $0')
> Was testing: `exec' with -l option
> ./A01grammar.ztst: test failed.

We don't use TIOCGETD (get terminal line discipline) directly, so either
this is well buried or it's coming from /bin/sh.

> ./V01zmodload.ztst: starting.
> *** /tmp/zsh.ztst.err.15666	Mon Apr 20 01:50:13 2009
> --- /tmp/zsh.ztst.terr.15666	Mon Apr 20 01:50:13 2009
> ***************
> *** 0 ****
> --- 1,2 ----
> + (eval):3: module `zsh/terminfo' has no such feature: `b:echoti': autoload cancelled
> + (eval):3: module `zsh/terminfo' has no such feature: `p:terminfo': autoload cancelled
> Test ./V01zmodload.ztst failed: error output differs from expected as shown above for:
>  for m in $mods
>  do
>    zmodload -i $m || mods[(r)$m]=()
>  done
> Was testing: Test loading of all compiled modules
> ./V01zmodload.ztst: test failed.

Somehow zsh/terminfo is getting compiled and linked in, implying the tests
in Src/Modules/terminfo.c are succeeding (link is "either" or "static"):

link='if test "x$ac_cv_func_tigetflag" = xyes -a "x$ac_cv_header_curses_h" = xyes ||
      test x$ac_cv_header_ncurses_h = xyes || test x$ac_cv_header_ncurses_ncurses_h = xyes ||
      test x$ac_cv_header_ncursesw_ncurses_h = xyes; then
          if test "x$zsh_cv_shared_tigetstr" = xyes; then
	      echo either
	  else
	      echo static
	  fi
      else
          echo no;
      fi
'

even though USE_TERMINFO_MODULE isn't being defined, i.e.

#if defined(HAVE_TIGETFLAG) && defined(ZSH_HAVE_CURSES_H)
# define USE_TERMINFO_MODULE 1
#else
# undef USE_TERMINFO_MODULE
#endif

HAVE_TIGETFLAG or ZSH_HAVE_CURSES_H are not defined.

We should really test zsh_cv_path_curses_header in terminfo.mdd, though I
don't know if this is the problem.

Index: Src/Modules/terminfo.mdd
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Modules/terminfo.mdd,v
retrieving revision 1.13
diff -u -r1.13 terminfo.mdd
--- Src/Modules/terminfo.mdd	5 Dec 2007 15:21:23 -0000	1.13
+++ Src/Modules/terminfo.mdd	20 Apr 2009 16:44:37 -0000
@@ -1,8 +1,6 @@
 name=zsh/terminfo
 
-link='if test "x$ac_cv_func_tigetflag" = xyes -a "x$ac_cv_header_curses_h" = xyes ||
-      test x$ac_cv_header_ncurses_h = xyes || test x$ac_cv_header_ncurses_ncurses_h = xyes ||
-      test x$ac_cv_header_ncursesw_ncurses_h = xyes; then
+link='if test "x$ac_cv_func_tigetflag" = xyes -a x$zsh_cv_path_curses_header != xnone; then
           if test "x$zsh_cv_shared_tigetstr" = xyes; then
 	      echo either
 	  else


-- 
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

Re: zsh 4.3.9-dev-2, two test failures on NetBSD

[Paul Ackersviller]

On Mon, Apr 20, 2009 at 05:48:08PM +0100, Peter Stephenson wrote:
> On Mon, 20 Apr 2009 04:00:18 +0000
> We don't use TIOCGETD (get terminal line discipline) directly, so either
> this is well buried or it's coming from /bin/sh.

Thanks, you're right it was /bin/sh.  The system turns out to have a
stty in /etc/profile that's really intended only for interactive shells.
The test is for login shells, so that makes sense.  Could there be a way to
work around such breakage to /etc/profile?  I can't think of anything easy.

> Somehow zsh/terminfo is getting compiled and linked in, implying the tests
> in Src/Modules/terminfo.c are succeeding (link is "either" or "static"):
...
> We should really test zsh_cv_path_curses_header in terminfo.mdd, though I
> don't know if this is the problem.
> 

Apparently so, this eliminates my test failure by not compiling terminfo
at all.  I'll see if I can configure it back in as well.

Re: zsh 4.3.9-dev-2, two test failures on NetBSD

[Peter Stephenson]

On Mon, 20 Apr 2009 21:48:22 +0000
Paul Ackersviller <pda@sdf.lonestar.org> wrote:
> On Mon, Apr 20, 2009 at 05:48:08PM +0100, Peter Stephenson wrote:
> > On Mon, 20 Apr 2009 04:00:18 +0000
> > We don't use TIOCGETD (get terminal line discipline) directly, so either
> > this is well buried or it's coming from /bin/sh.
> 
> Thanks, you're right it was /bin/sh.  The system turns out to have a
> stty in /etc/profile that's really intended only for interactive shells.
> The test is for login shells, so that makes sense.  Could there be a way to
> work around such breakage to /etc/profile?  I can't think of anything easy.

It might help to use zsh instead of sh here; zsh respects -f for login
shells and all we need to know is whether the "-" is at the start of the
line.

Index: Test/A01grammar.ztst
===================================================================
RCS file: /cvsroot/zsh/zsh/Test/A01grammar.ztst,v
retrieving revision 1.18
diff -u -r1.18 A01grammar.ztst
--- Test/A01grammar.ztst	27 Feb 2009 10:23:58 -0000	1.18
+++ Test/A01grammar.ztst	21 Apr 2009 09:33:43 -0000
@@ -62,9 +62,9 @@
   (exec /bin/sh; echo bar)
 0:`exec' precommand modifier
 
-  (exec -l /bin/sh -c 'echo $0')
+  (exec -l $ZTST_testdir/../Src/zsh -fc 'echo $0' | sed 's%/.*/%%' )
 0:`exec' with -l option
->-/bin/sh
+>-zsh
 
   (exec -a /bin/SPLATTER /bin/sh -c 'echo $0')
 0:`exec' with -a option


-- 
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

Re: zsh 4.3.9-dev-2, two test failures on NetBSD

[Paul Ackersviller]

On Tue, Apr 21, 2009 at 10:35:37AM +0100, Peter Stephenson wrote:
> On Mon, 20 Apr 2009 21:48:22 +0000
> Paul Ackersviller <pda@sdf.lonestar.org> wrote:
> > Thanks, you're right it was /bin/sh.  The system turns out to have a
> > stty in /etc/profile that's really intended only for interactive shells.
> > The test is for login shells, so that makes sense.  Could there be a way to
> > work around such breakage to /etc/profile?  I can't think of anything easy.
> 
> It might help to use zsh instead of sh here; zsh respects -f for login
> shells and all we need to know is whether the "-" is at the start of the
> line.

Perfect, that does get around the problem.  I wasn't even aware -f had
that effect!