It turns out the (#qN) doesn't work if you don't have EXTENDED_GLOB set, in
which case it should just be (N).  I'll see if I can find a better solution.

The way I chose is to put the command "emulate -L ksh" before sourcing
/etc/profile.d/*.sh.  This is probably a good idea anyway, since stuff in
/etc/profile.d is designed for bash, not zsh, and this will turn off any
incompatible options and make zsh behave more like standard sh/ksh/bash.

Note that it should be done inside its own block, so that the emulation only
applies for sourcing those scripts.

The final code looks like:
{
    emulate -L ksh
    if [ "x$SHLVL" != "x1" ]; then # We're not a login shell
        for i in /etc/profile.d/*.sh; do
            if [ -r "$i" ]; then
                . $i
            fi
        done
        unset i
    fi
}

See the URL for the long explanation of differences between zsh and sh (and zsh
and bash), and how the emulate command works.

I've been using emulate -L ksh for a couple of years in my .zshenv when sourcing
.shrc in my home directory, and it works quite well.

> /etc/zshrc assumes that the NOMATCH shell option is unset.

> If it is (e.g. in ~/.zshenv) and there are no files in /etc/profile.d, the
> following error is printed.

 But ~/.zshrc is sourced _after_ /etc/zshrc ... what is your actual problem?

 Wait, are you are putting interactive commands in ~/.zprofile? ... if so don't
do that.

The order is /etc/zshenv, ~/.zshenv, /etc/zprofile, ~/.zprofile, /etc/zshrc,
~/.zshrc, /etc/zlogin, ~/.zlogin.  You can verify this in the zsh man page.

My ~/.zshenv sets all my options that I want to use everywhere (scripts,
console, etc.).

My ~/.zprofile sets stuff that I only want to run at login time, such as running
my ssh agent and setting up a script to run at logout time.

My ~/.zshrc sets up my interactive settings, such as aliases and prompt.

There is nothing wrong with this set up.  The problem is that /etc/zshrc assumes
certain conditions that may not be true.  It can be fixed by running
/etc/profile.d/*.sh in sh compatibility mode, which is a good idea anyway, since
those scripts were written for bash not zsh.

 Well I would classify the NOMATCH option as an interactive thing, but I'll do
the above fix for rawhide/Fedora 9.

You're right, it's probably not very common, but it's very handy to be able to
have exactly the same environment for scripts as at the console.  It means I can
be sure something I run from the console works the same when I put it in a
script and run it from cron or wherever.

I should note that the NOMATCH option is in fact the default in zsh, it's just
you won't see this behavior unless your /etc/profile.d contains no scripts.

Also, this fix should fix a whole class of other problems that could occur in
scripts under /etc/profile.d, including wildcard matching ("globbing"), arrays,
and word splitting, as the URL explains.

Thanks very much for looking at it.

Sorry, the fix I suggested only works if it's a separate function (which limits
the scope of the sh emulate to that function, and reverts to the standard/user
zsh options after the function exits).

It should be something like:
_source_profile.d()
{
    emulate -L ksh
    ...
}

_source_profile.d

This should be used in /etc/zprofile and /etc/zshrc, seeing as
the/etc/profile.d/*.sh scripts are sourced in both of those files.

Ok, 4.3.4-7.fc9 should be in rawhide soon and contains the fix for both files
using functions.

Guys, I can't reopen this but hopefully you still get my message:

This Bug 430665 is almost a word-by-word duplicate of Bug 225454 but it is not
marked so. Also, the Bug 225454 includes a solution blessed in the Z shell
mailing lists and it is also much more compact (I would even say more elegant,
too) than the one applied to latest RPM in rawhide.

Please have a look at:

https://bugzilla.redhat.com/show_bug.cgi?id=225454#c6

Thanks.

Your proposal is to remove /etc/zshrc.  The Red Hat maintainers want this kept for weird cases such as 
the user invokes a different terminal type and /etc/profile.d contains something specific for that terminal.  
(See bug 430827 for an example.)

So the only difference I can see is the Red Hat one has some ugly SHLVL tests, but they're there for a 
reason.

 Also note that this close to Fedora 9 there's roughly 0% chance of changing zsh
unless there's a significant bug.
 Post Fedora 9 we can look at using "sh" emulation instead of "ksh", if that is
better for some cases.

Thanks for your replies. I've been waiting for a long to get these fixed so I'm
more than happy if we can get back to this after F9 is out.

*** Bug 225454 has been marked as a duplicate of this bug. ***

This fix has broken reading of my user .zshrc. For some reason after /etc/zshrc
has finished zsh no longer reads my .zshrc file. I reverted just this change out
of /etc/{zprofile,zshrc} and everything started working normally again.

Here is the last few lines of zsh -lx with current rawhide:

: /etc/zprofile:22 00:23:38 {$SECONDS} ; unset -f _src_etc_profile
: /etc/zshrc:11 00:23:38 {$SECONDS} ; PROMPT='[%n@%m]%~%# ' 
: /etc/zshrc:16 00:23:38 {$SECONDS} ; bindkey ' ' magic-space
: /etc/zshrc:35 00:23:38 {$SECONDS} ; _src_etc_profile_d
: _src_etc_profile_d:4 00:23:38 {$SECONDS} ; emulate -L ksh
: _src_etc_profile_d:8 00:23:38 {$SECONDS} ; [[ ! -o login ]]
: /etc/zshrc:37 00:23:38 {$SECONDS} ; unset -f _src_etc_profile_d

As you can see, /etc/zshrc has finished but .zshrc has not been executed. With
this change reverted those few lines of zsh -lx become:

: /etc/zshrc:11 00:35:11 {$SECONDS} ; PROMPT='[%n@%m]%~%# ' 
: /etc/zshrc:16 00:35:11 {$SECONDS} ; bindkey ' ' magic-space
: /etc/zshrc:19 00:35:11 {$SECONDS} ; [[ ! -o login ]]
: /home/martin/.zsh/.zshrc:7 00:35:11 {$SECONDS} ; ZFILE=zshrc 
: /home/martin/.zsh/.zshrc:8 00:35:11 {$SECONDS} ; [[ -n '' ]]
: /home/martin/.zsh/.zshrc:17 00:35:11 {$SECONDS} ; _init_zshrc

Note that I have ZDOTDIR set to ~/.zsh, I don't know if that is a factor in the bug.

Martin,

1) can you give instructions to fully replicate your environment?
2) do you have anything special in /etc/profile.d? what if you move those files aside?
3) are all the files you source executable? are you relying on them being in PATH or do you specify the 
full path?
4) can you narrow down what part is causing the problem? what if you only revert zshrc? only zprofile?  
what if you start an interactive, non-login shell?
5) where are you setting ZDOTDIR?
6) what is the full list of files it touches before and after the change?
7) what are the differences in the output of "set" before and after the new code?
8) is it easy for you to replicate the problem without ZDOTDIR? can you produce a minimal test case?
9) is there anything else peculiar to your environment, like is the -e (exit on error) option enabled, or 
do you already use "emulate" in some way?

A couple of other ideas:
- try "unsetopt GLOBAL_RCS" in your .zshenv (to isolate that it's the /etc/ scripts causing the problem, 
and also as a possible workaround)
- do you already set (or unset) GLOBAL_RCS or RCS at any point?
- do you ever unset LOCAL_OPTIONS or do anything else that might change the behavior of functions and 
local variables?

I've managed to reproduce this with a blank config, is is due to breakage of
ZDOTDIR.

% adduser zshtest
% passwd zshtest
% su - zshtest
% echo 'export ZDOTDIR=$HOME/.zsh' > ~/.zshenv 
% echo "echo In .zsh/.zshrc" > ~/.zsh/.zshrc 
% echo "echo In .zshrc" > ~/.zshrc

With the current rawhide:
% zsh
In .zshrc

This is wrong. With the change in this bug reverted:

% zsh
In .zsh/.zshrc

This is correct. Emptying out /etc/profile.d didn't make any difference.

Sorry, don't forget -s to adduser:
adduser -s /bin/zsh zshtest

I don't have access to my Linux system right now.

Is ZDOTDIR still set at the end of /etc/zprofile?  /etc/zshrc?  If not, can you figure out which line is 
unsetting it?

Thanks for the info!

Created attachment 303911 [details]
ZSH startup with bug

Created attachment 303912 [details]
ZSH startup without bug

Nothing stands out as an obvious cause.

Added some echos, ZDOTDIR is not getting unset at all, for some reason it is
just being ignored. Just removing the emulate commands from inside the functions
in /etc/z* is sufficient to fix it. Looks like the effect of emulate is not
localised to within the function like it should be.

It's actually unsetting then restoring ZDOTDIR.  That is expected.  I'm not sure
why echo doesn't show that.

$ cat .zshrc
echo "In .zshrc"
typeset -p ZDOTDIR
typeset -p GLOBAL_RCS
typeset -p RCS

emulate_test()
{
       emulate -L ksh

       echo "In emulate_test"
       typeset -p ZDOTDIR
       typeset -p GLOBAL_RCS
       typeset -p RCS
}

emulate_test

echo "Out .zshrc"
typeset -p ZDOTDIR
typeset -p GLOBAL_RCS
typeset -p RCS

# su - zshtest
In .zshrc
typeset -x ZDOTDIR=/home/zshtest/.zsh
typeset GLOBAL_RCS=''
typeset RCS=''
In emulate_test
typeset ZDOTDIR=''
typeset GLOBAL_RCS=''
typeset RCS=''
Out .zshrc
typeset -x ZDOTDIR=/home/zshtest/.zsh
typeset GLOBAL_RCS=''
typeset RCS=''

It doesn't yet explain the problem, but it's something for me to look into. 
Perhaps zsh does something weird if it gets set then unset.

Given that the original bug has been present for a long time, and that this fix
is causing regressions maybe it is worth removing the fix for the F9 release and
finding a complete solution after the release has been made?

 So putting:

# Minor bug in zsh, with regard to emulate -L see rhbz#430665
_src_noop()
{
    emulate -L zsh
}
_src_noop

unset -f _src_noop

...at the end of the init scripts solve the testcase, for me. It seems a little
too much like magic, but putting debugging code within zsh shows that ZDOTDIR
only gets reset again when the next file is sourced (i.e. emulate is implicitly
called).

 Martin does this fix it for you?
 If so I'll make sure the new rcfiles are in updates-testing when Fed-9 goes GA
(but probably not in updates itself for a couple of weeks after).

Yes that does fix it.

Changing version to '9' as part of upcoming Fedora 9 GA.
More information and reason for this action is here:
http://fedoraproject.org/wiki/BugZappers/HouseKeeping

zsh-4.3.4-8.fc9 has been submitted as an update for Fedora 9

zsh-4.3.4-8.fc9 has been pushed to the Fedora 9 testing repository.  If problems still persist, please make note of it in this bug report.
 If you want to test the update, you can install it with 
 su -c 'yum --enablerepo=updates-testing update zsh'.  You can provide feedback for this update here: http://admin.fedoraproject.org/updates/F9/FEDORA-2008-4099

zsh-4.3.4-8.fc9 has been pushed to the Fedora 9 stable repository.  If problems still persist, please make note of it in this bug report.

I was able to reproduce the emulation problem. I reported this upstream at:

http://www.zsh.org/mla/users/2008/msg00781.html

And, amazingly, within 24 hours on a Saturday/Sunday period a patch was posted
and committed:

http://www.zsh.org/mla/users/2008/msg00792.html
http://zsh.cvs.sourceforge.net/zsh/zsh/ChangeLog?revision=1.4264&view=markup

I think it would make sense to remove the added workaround when upgrading zsh sources to a version that includes the fix.

The unneeded workaround is still present in zsh-4.3.9-4.fc11.i586.

This message is a reminder that Fedora 9 is nearing its end of life.
Approximately 30 (thirty) days from now Fedora will stop maintaining
and issuing updates for Fedora 9.  It is Fedora's policy to close all
bug reports from releases that are no longer maintained.  At that time
this bug will be closed as WONTFIX if it remains open with a Fedora 
'version' of '9'.

Package Maintainer: If you wish for this bug to remain open because you
plan to fix it in a currently maintained version, simply change the 'version' 
to a later Fedora version prior to Fedora 9's end of life.

Bug Reporter: Thank you for reporting this issue and we are sorry that 
we may not be able to fix it before Fedora 9 is end of life.  If you 
would still like to see this bug fixed and are able to reproduce it 
against a later version of Fedora please change the 'version' of this 
bug to the applicable version.  If you are unable to change the version, 
please add a comment here and someone will do it for you.

Although we aim to fix as many bugs as possible during every release's 
lifetime, sometimes those efforts are overtaken by events.  Often a 
more recent Fedora release includes newer upstream software that fixes 
bugs or makes them obsolete.

The process we are following is described here: 
http://fedoraproject.org/wiki/BugZappers/HouseKeeping

Fedora 9 changed to end-of-life (EOL) status on 2009-07-10. Fedora 9 is 
no longer maintained, which means that it will not receive any further 
security or bug fix updates. As a result we are closing this bug.

If you can reproduce this bug against a currently maintained version of 
Fedora please feel free to reopen this bug against that version.

Thank you for reporting this bug and we are sorry it could not be fixed.