MIME-Version: 1.0
In-Reply-To: <874n4hy8sa.fsf@nikula.org>
References:
 <CANbPJF=vD8VAhGKT_3udssvApbD_W644j=4o2H8h11xZjqPphA@mail.gmail.com>
	<87a9e9yeta.fsf@nikula.org> <874n4hy8sa.fsf@nikula.org>
Date: Sun, 2 Feb 2014 15:33:26 +0100
Message-ID:
 <CANbPJF=3+KpND6Fmhwuwer4hPBC6SKHM3C_dXr3xRzg8svk0uQ@mail.gmail.com>
Subject: Re: bash completion error with 0.17
From: Anand Buddhdev <arhbkb@gmail.com>
To: Jani Nikula <jani@nikula.org>
Content-Type: multipart/alternative; boundary=f46d0442834c49b9a104f16d4afd
Cc: notmuch@notmuchmail.org
Precedence: list

--f46d0442834c49b9a104f16d4afd
Content-Type: text/plain; charset=ISO-8859-1

Thanks Jani. I've just looked at the bash-completion formula on Homebrew.
It says that bash-completion 2.0 and above needs bash4, but the current
version of OSX, 10.9, still has bash 3.2, which is why, at least under
Homebrew, we can't update bash-completion to the latest version.

I think your patch probably makes sense. Once a user updates to a newer
bash-completion, it will just begin working, while for users like me, the
error message goes away.


On 2 February 2014 13:54, Jani Nikula <jani@nikula.org> wrote:

> On Sun, 02 Feb 2014, Jani Nikula <jani@nikula.org> wrote:
> > The trivial "fix" patch would likely just check for the existence of
> > _init_completion, and refuse to complete without it. For anything
> > fancier, patches welcome...
>
> The patch below would get rid of the error message (and completion, for
> that matter) when _init_completion is not present.
>
> I'm not sure whether this is the way to go, or should we rather check
> for a recent enough bash-completion during configure, or
> recommend/suggest it, or what. The error message does tell us what's
> wrong, although it's annoying for the user (probably more so than just
> not completing).
>
> BR,
> Jani.
>
>
>
> diff --git a/completion/notmuch-completion.bash
> b/completion/notmuch-completion.bash
> index 0f13204..0c457a8 100644
> --- a/completion/notmuch-completion.bash
> +++ b/completion/notmuch-completion.bash
> @@ -384,6 +384,10 @@ _notmuch()
>  {
>      local _notmuch_commands="compact config count dump help insert new
> reply restore search setup show tag"
>      local arg cur prev words cword split
> +
> +    # require bash-completion with _init_completion
> +    type -t _init_completion >/dev/null 2>&1 || return
> +
>      _init_completion || return
>
>      COMPREPLY=()
>

--f46d0442834c49b9a104f16d4afd
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Thanks Jani. I&#39;ve just looked at the bash-completion f=
ormula on Homebrew. It says that bash-completion 2.0 and above needs bash4,=
 but the current version of OSX, 10.9, still has bash 3.2, which is why, at=
 least under Homebrew, we can&#39;t update bash-completion to the latest ve=
rsion.<div>
<br></div><div>I think your patch probably makes sense. Once a user updates=
 to a newer bash-completion, it will just begin working, while for users li=
ke me, the error message goes away.</div></div><div class=3D"gmail_extra">
<br><br><div class=3D"gmail_quote">On 2 February 2014 13:54, Jani Nikula <s=
pan dir=3D"ltr">&lt;<a href=3D"mailto:jani@nikula.org" target=3D"_blank">ja=
ni@nikula.org</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" st=
yle=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class=3D"im">On Sun, 02 Feb 2014, Jani Nikula &lt;<a href=3D"mailto:ja=
ni@nikula.org">jani@nikula.org</a>&gt; wrote:<br>
&gt; The trivial &quot;fix&quot; patch would likely just check for the exis=
tence of<br>
&gt; _init_completion, and refuse to complete without it. For anything<br>
&gt; fancier, patches welcome...<br>
<br>
</div>The patch below would get rid of the error message (and completion, f=
or<br>
that matter) when _init_completion is not present.<br>
<br>
I&#39;m not sure whether this is the way to go, or should we rather check<b=
r>
for a recent enough bash-completion during configure, or<br>
recommend/suggest it, or what. The error message does tell us what&#39;s<br=
>
wrong, although it&#39;s annoying for the user (probably more so than just<=
br>
not completing).<br>
<br>
BR,<br>
Jani.<br>
<br>
<br>
<br>
diff --git a/completion/notmuch-completion.bash b/completion/notmuch-comple=
tion.bash<br>
index 0f13204..0c457a8 100644<br>
--- a/completion/notmuch-completion.bash<br>
+++ b/completion/notmuch-completion.bash<br>
@@ -384,6 +384,10 @@ _notmuch()<br>
=A0{<br>
=A0 =A0 =A0local _notmuch_commands=3D&quot;compact config count dump help i=
nsert new reply restore search setup show tag&quot;<br>
=A0 =A0 =A0local arg cur prev words cword split<br>
+<br>
+ =A0 =A0# require bash-completion with _init_completion<br>
+ =A0 =A0type -t _init_completion &gt;/dev/null 2&gt;&amp;1 || return<br>
+<br>
=A0 =A0 =A0_init_completion || return<br>
<br>
=A0 =A0 =A0COMPREPLY=3D()<br>
</blockquote></div><br></div>

--f46d0442834c49b9a104f16d4afd--