Date: Sun, 11 Dec 2011 19:31:03 -0500
From: Austin Clements <amdragon@MIT.EDU>
To: Jani Nikula <jani@nikula.org>
Subject: Re: [PATCH] emacs: support "notmuch new" as a notmuch-poll-script
Message-ID: <20111212003103.GI2760@mit.edu>
References: <1323640100-18326-1-git-send-email-jani@nikula.org>
	<87wra26a5u.fsf@gmail.com> <20111211225822.GH2760@mit.edu>
	<CAB+hUn-ADRtZRCq5Ad6H6R2L246EBR8hOSsk4B2yeRsH4kN16w@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To:
 <CAB+hUn-ADRtZRCq5Ad6H6R2L246EBR8hOSsk4B2yeRsH4kN16w@mail.gmail.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
Cc: notmuch@notmuchmail.org
Precedence: list

Quoth Jani Nikula on Dec 12 at  1:10 am:
>    On Dec 12, 2011 12:56 AM, "Austin Clements" <[1]amdragon@mit.edu> wrote:
>    >
>    > Quoth Dmitry Kurochkin on Dec 12 at  2:00 am:
>    > > Hi Jani.
>    > >
>    > > On Sun, 11 Dec 2011 23:48:20 +0200, Jani Nikula <[2]jani@nikula.org>
>    wrote:
>    > > > Let notmuch-poll-script be a function as well as a string. Make
>    default
>    > > > value nil instead of an empty string, but allow "" for backwards
>    > > > compatibility. Add a notmuch poll function to call "notmuch new"
>    using the
>    > > > configured notmuch-command.
>    > > >
>    > > > This allows taking better advantage of the "notmuch new" hooks from
>    emacs
>    > > > without intermediate scripts.
>    > > >
>    > >
>    > > I was just thinking about working on this myself :)
>    > >
>    > > I think a better solution would be to allow running a command with
>    > > arguments.  Creating a elisp function just to run a command with some
>    > > parameters feels wrong.  This way we would have to add another
>    function
>    > > each time we want to add another argument.
>    >
>    > This seems a little awkward to me, too, though perhaps it's the best
>    > way.  Other approaches to consider include accepting a list for
>    > notmuch-poll-script (e.g., ("notmuch" "new")) or leaving it as a
>    > string but treating it as a shell command so "notmuch new" would Just
>    > Work.  Personally, I think the latter is the most intuitive, but it
>    > would be worth looking at how other customizable external commands are
>    > done in Emacs.
>    >
>    > A function seems powerful, but also like overkill.  Can you give a use
>    > case for a function that wouldn't be more easily solved by one of the
>    > above approaches?
> 
>    The only reason I had for using a function was running notmuch using
>    notmuch-command. Any ideas how to do that with the Just Works approach?

Oh, I see.  I'd missed that.

So here's another idea, prefaced with a rant.

It's bothered me for a long time that notmuch-emacs didn't just know
by default how to check for new mail.  What MUA doesn't know how to
check for new mail?  Why does a new user of notmuch have to tell it
how to check for new mail?  Of course, this *had* to be configured
before because everyone had their own way of checking for new mail.
Hooks eliminate this unnecessary flexibility and make "notmuch new"
the one true way to check for new mail---as it ought to be---and in
turn make the notmuch-poll-script variable obsolete.

So, what about changing the default "" setting of notmuch-poll-script
from meaning "do nothing and be useless" to meaning "run notmuch new
(using notmuch-command)"?  It will then automatically do the right
thing for new users, while still being backward-compatible and
allowing an escape hatch for bizarre situations.