patchgit-completion: workaround zsh COMPREPLY bug

zsh adds a backslash (foo\ ) for each item in the COMPREPLY array if IFS doesn't contain spaces. This issue has been reported[1], but there is no solution yet.

This wasn't a problem due to another bug[2], which was fixed in zsh version 4.3.12. After this change, 'git checkout ma<tab>' would resolve to 'git checkout master\ '.

Aditionally, the introduction of __gitcomp_nl in commit a31e626 (completion: optimize refs completion) in git also made the problem apparent, as Matthieu Moy reported.

The simplest and most generic solution is to hide all the changes we do to IFS, so that "foo \nbar " is recognized by zsh as "foo bar". This works on versions of git before and after the introduction of __gitcomp_nl (a31e626), and versions of zsh before and after 4.3.12.

Once zsh is fixed, we should conditionally disable this workaround to have the same benefits as bash users.

[1] http://www.zsh.org/mla/workers/2012/msg00053.html [2] http://zsh.git.sourceforge.net/git/gitweb.cgi?p=zsh/zsh;a=commitdiff;h=2e25dfb8fd38dbef0a306282ffab1d343ce3ad8d

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
---
 contrib/completion/git-completion.bash |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)

Felipe Contreras <felipe.contreras@gmail.com> writes:

[...]

No time to test right now, but is this not going to

1) leave IFS as hidden even outside the completion script, possibly
affecting unrelated scripts that would need to set IFS as local and keep
its special effect?

2) break cases where strings are to be split on \n only (e.g. see
"foo bar\nboz" as three possible completions "foo", "bar", "boz" instead
of "foo bar" and "boz"?

On Wed, Jan 25, 2012 at 9:41 AM, Matthieu Moy <Matthieu.Moy@grenoble-inp.fr> wrote:

What special effect?

Unrelated scripts can still set IFS as local.

You can test this:

--- #!/bin/sh

if [[ -n ${ZSH_VERSION-} ]]; then
    autoload -U +X bashcompinit && bashcompinit
fi

_foo()
{
    typeset -h IFS
    local cur="${COMP_WORDS[COMP_CWORD]}"
    local IFS=$'\n'
    w[0]='first '
    w[1]='second and space '
    w[2]='third\\ quoted\\ space '
    ./test1 >> /tmp/t1.txt
    COMPREPLY=( $(compgen -W "${w[*]}" -- $cur) )
}
complete -o nospace -F _foo foo
---

test1: --- #!/bin/sh

test1() {
	local IFS=$'\n'
	w=( $(echo -e "foo \nbar"))
	echo "test1: 0:'${w[0]}' 1:'${w[1]}'"
}

test1

w=( $(echo -e "foo \nbar")) echo "test2: 0:'${w[0]}' 1:'${w[1]}'" ---

The result in /tmp/t1.txt would be: test1: 0:'foo ' 1:'bar' test2: 0:'foo' 1:'bar'

Regardless of whether you use typeset -h or not.

Those cases are already broken, which is what I reported to the zsh mailing list. You would get "foo\ bar" and "boz", and that's after 4.3.12; before, compgen -W would still break the completions in 3, because they did 'results+=( $words )', instead of 'results+=( "$words" )'.

Cheers.

Felipe Contreras <felipe.contreras@gmail.com> writes:

OK, I missed the fact that typeset -h had only local effect.

Makes sense.

I still have a minor comment: maybe part of your commit message could go to a comment in the code as well, in particular the "Once zsh is fixed" part, to help future contributors to actually disable the workaround when possible in the future.

Tested-by: Matthieu Moy <Matthieu.Moy@imag.fr>

On Thu, Jan 26, 2012 at 12:02 AM, Matthieu Moy <Matthieu.Moy@grenoble-inp.fr> wrote:

FTR, I've fixed all the issues in zsh's bash completion:

http://thread.gmane.org/gmane.comp.shells.zsh.devel/24290

Now I don't need the workaround and it works just like in bash :) However, zsh devs seem very quiet.

Cheers.

Felipe Contreras <felipe.contreras@gmail.com> writes:

That 2e25dfb8 only says:

    Rocky Bernstein: 29135 (plus tweaks): compgen -W in bash completion

without any explanation, which is not very useful.

Do you have a bug tracker ID or something for [1] above, with which I can amend the patch as Matthieu suggests?

On Fri, Jan 27, 2012 at 12:00 AM, Junio C Hamano <gitster@pobox.com> wrote:

Yeah, they development practices leaves a lot to be desired.

I don't think there's something like that, but here's the original discussion:

http://thread.gmane.org/gmane.comp.shells.zsh.devel/22541