From: Jani Nikula <jani@nikula.org>
To: Tomi Ollila <tomi.ollila@iki.fi>, John Lenz <lenz@math.uic.edu>,
	notmuch@notmuchmail.org
Subject: Re: cli: add --include-html option to notmuch show
In-Reply-To: <m2wqni7swv.fsf@guru.guru-group.fi>
References: <notmuch-web-1372724382.450184839@www.wuzzeb.org>
	<m27ggj1x29.fsf@guru.guru-group.fi>
	<notmuch-web-1374719771.1588310986@www.wuzzeb.org>
	<m2zjsxs0g1.fsf@guru.guru-group.fi>
	<notmuch-web-1376754748.336647450@www.wuzzeb.org>
	<87k3jjutpf.fsf@nikula.org> <m2wqni7swv.fsf@guru.guru-group.fi>
User-Agent: Notmuch/0.16+10~g791e68f (http://notmuchmail.org) Emacs/24.3.1
	(x86_64-pc-linux-gnu)
Date: Sat, 24 Aug 2013 13:59:20 +0300
Message-ID: <878uzrnylz.fsf@nikula.org>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: list

On Sun, 18 Aug 2013, Tomi Ollila <tomi.ollila@iki.fi> wrote:
> Now, if we have --include-html should it be like that or 
> --include-html=(true|false). Currently we have both cases, adding
> --verify, --decrypt, --create-folder, --batch, -no-hooks to the
> set... I cannot get a clear opinion (without wast^H^H^H^H spending
> excessive amount of time figuring these out) how this should be,
> therefore I'm inclined to the opinion that
>
> the current patch from John with simple --include-html could be applied,
> and in the future (if it is of anyone's interest) we update the parser
> allowing boolean --arg equal --arg=true. Then it is just how we decide
> to document these...

The argument parser we have allows NOTMUCH_OPT_BOOLEAN options to be
specified as --foo=(true|false) or simply --foo (for true). It is
already now just a matter of documentation, and I'm sure we're not
consistent.

We also have things like --no-hooks in notmuch new. I added that, and in
retrospect, that should be just --hooks=(true|false). I guess we left it
like that because the default is true. Perhaps we should amend the
argument parser to look for boolean option --foo if it encounters a
--no-foo option.

Why notmuch show has --exclude=(true|false) as a keyword option I don't
quite grasp; the --entire-thread option at least has a special
default. It might help some options if the parser had a way to tell if
it's seen some option or not.

Of course, none of this is really relevant to the patch in question!

BR,
Jani.