PATCH: Fix ZDOTDIR during "make check" for static builds

PATCH: Fix ZDOTDIR during "make check" for static builds

[Bart Schaefer]

When I made my earlier patch for this (12039), I forgot that Test/Modules
doesn't get created when the build is static.

This points ZDOTDIR into the Test directory itself (in the build tree, if
that's separate from the source), and therefore also has to make sure that
the .zcompdump file is removed at the end.

Index: Test/Makefile.in
===================================================================
@@ -48,7 +48,7 @@
 	for f in $(sdir)/$(TESTNUM)*.ztst; do \
 	  $(dir_top)/Src/zsh -f $(sdir)/ztst.zsh $$f; \
 	done
-	rm -rf Modules
+	rm -rf Modules .zcompdump
 
 # ========== DEPENDENCIES FOR CLEANUP ==========
 
Index: Test/comptest
===================================================================
@@ -25,7 +25,7 @@
   }
 
   comptesteval \
-"export ZDOTDIR=$ZTST_testdir/Modules" \
+"export ZDOTDIR=$ZTST_testdir" \
 "module_path=( $module_path )" \
 "fpath=( $fpath )" \
 'LISTMAX=10000000

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net   

Test hanger disappeared, but ... RE: PATCH: Fix ZDOTDIR during "make check" for static builds

[Andrej Borsenkow]

>
> When I made my earlier patch for this (12039), I forgot that
> Test/Modules
> doesn't get created when the build is static.
>
> This points ZDOTDIR into the Test directory itself (in the
> build tree, if
> that's separate from the source), and therefore also has to
> make sure that
> the .zcompdump file is removed at the end.
>

May be, it is something different - but now completion tests do not hang
anymore. I run them on Unix without any problems; but on Cygwin I get
misterious errors:

mw1g017@MW1G17C:/usr/build/zsh/Test% /usr/build/zsh/Src/zsh -f
/usr/src/zsh/Test
/ztst.zsh /usr/src/zsh/Test/53completion.ztst
*** /tmp/zsh.ztst.out.1007      Tue Jul  4 12:24:39 2000
--- /tmp/zsh.ztst.tout.1007     Tue Jul  4 12:24:49 2000
***************
*** 1,12 ****
- line: {: }{}
  DESCRIPTION:{file}
  DI:{dir1}
  DI:{dir2}
  FI:{file1}
  FI:{file2}
- line: {: dir1/}{}
- line: {: dir2/}{}
- line: {: file1}{}
- line: {: file2}{}
- line: {: dir1/}{}
- line: {: dir2/}{}
--- 1,5 ----
Test /usr/src/zsh/Test/53completion.ztst failed: output differs from
expected as
 shown above for:
  comptest $': \t\t\t\t\t\t\t'
Was testing: directories and files
/usr/src/zsh/Test/53completion.ztst: test failed.
rm: /usr/build/zsh/Test/comp.tmp: Permission denied
mw1g017@MW1G17C:/usr/build/zsh/Test% print $CYGWIN
title strip_title tty

I thought, it may have something to do with CYGWIN setting (tty/notty) -
but the error happens with both settings.

The "permission denied" message most probably comes, because some child
has comp.tmp open at the time comptest tries to delete it.

It all smells much like some tty emulation problem. Sigh.

-andrej

PATCH: compdump (Re: Test hanger disappeared, but ...)

[Bart Schaefer]

On Jul 4, 12:29pm, Andrej Borsenkow wrote:
} 
} > When I made my earlier patch for this (12039), I forgot that
} > Test/Modules doesn't get created when the build is static.
} 
} May be, it is something different - but now completion tests do not hang
} anymore.

It looks like compdump failure causes compinit to abort completely when
ZDOTDIR points to a nonexistent location.  The redirection on line 34
prints "no such file or directory" and then everything stops.

We seem to be running into a lot of things that cause zsh to abort all
the way back to top level.  Do other shells give up so thoroughly in
these situations?

Index: Completion/Core/compdump
===================================================================
@@ -21,6 +21,8 @@
 _d_file=${_comp_dumpfile-${0:h}/compinit.dump}.$HOST.$$
 [[ $_d_file = //* ]] && _d_file=${_d_file[2,-1]}
 
+[[ -w ${_d_file:h} ]] || return 1
+
 _d_files=( ${^~fpath:/.}/^([^_]*|*~|*.zwc)(N) )
 
 if [[ -n "$_comp_secure" ]]; then

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net   

RE: PATCH: Fix ZDOTDIR during "make check" for static builds

[Andrej Borsenkow]

>
> This points ZDOTDIR into the Test directory itself (in the
> build tree, if
> that's separate from the source), and therefore also has to
> make sure that
> the .zcompdump file is removed at the end.
>

If the very last test fails,  neither .zcompdump nor Modules directory
are removed. Obviously, for loop returns status of last executed
command. I forgot, how to force make to ignore it :-(

-andrej

Re: PATCH: Fix ZDOTDIR during "make check" for static builds

[Bart Schaefer]

On Jul 4,  9:01pm, Andrej Borsenkow wrote:
} Subject: RE: PATCH: Fix ZDOTDIR during "make check" for static builds
}
} If the very last test fails,  neither .zcompdump nor Modules directory
} are removed.  [...]  I forgot, how to force make to ignore it :-(

Like this.

Index: Test/Makefile.in
===================================================================
@@ -45,7 +45,7 @@
 	  cd $(dir_top) && \
 	  $(MAKE) MODDIR=`pwd`/$(subdir)/Modules install.modules > /dev/null; \
 	fi
-	for f in $(sdir)/$(TESTNUM)*.ztst; do \
+	-for f in $(sdir)/$(TESTNUM)*.ztst; do \
 	  $(dir_top)/Src/zsh -f $(sdir)/ztst.zsh $$f; \
 	done
 	rm -rf Modules .zcompdump

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net   

Re: PATCH: compdump (Re: Test hanger disappeared, but ...)

[Peter Stephenson]

Sven wrote:
> That's interesting... For example, with
> 
>   while true; do echo > unwritable; sleep 1; done
> 
> zsh stops after the first attempt. bash and ksh don't stop and the
> /bin/sh here stops after the first attempt, too.
> 
> With
> 
>   while true; do echo $(for); sleep 1; done
> 
> zsh stops (and prints two error messages!), bash does not stop, but
> ksh and /bin/sh do.

These are all perfectly arguable.  In the second case, it's only because
parsing is done before the fork that the shell actually gives up.  If you
attempted to write to an unwritable file in the $(...) it wouldn't stop.  I
don't have any conclusion.

-- 
Peter Stephenson <pws@cambridgesiliconradio.com>
Cambridge Silicon Radio, Unit 300, Science Park, Milton Road,
Cambridge, CB4 0XL, UK                          Tel: +44 (0)1223 392070

Re: PATCH: compdump (Re: Test hanger disappeared, but ...)

[Sven Wischnowsky]

Bart Schaefer wrote:

> ...
> 
> We seem to be running into a lot of things that cause zsh to abort all
> the way back to top level.  Do other shells give up so thoroughly in
> these situations?

That's interesting... For example, with

  while true; do echo > unwritable; sleep 1; done

zsh stops after the first attempt. bash and ksh don't stop and the
/bin/sh here stops after the first attempt, too.

With

  while true; do echo $(for); sleep 1; done

zsh stops (and prints two error messages!), bash does not stop, but
ksh and /bin/sh do.

Bye
 Sven


--
Sven Wischnowsky                         wischnow@informatik.hu-berlin.de