Re: [PATCH 3/8] tag: Change misleading --list <pattern> documentation

From: Jeff King <peff@peff.net>
To: "Ævar Arnfjörð Bjarmason" <avarab@gmail.com>
Cc: Junio C Hamano <gitster@pobox.com>,
	Git Mailing List <git@vger.kernel.org>,
	Lars Hjemli <hjemli@gmail.com>,
	Christian Couder <christian.couder@gmail.com>,
	Carlos Rica <jasampler@gmail.com>,
	Samuel Tardieu <sam@rfc1149.net>,
	Tom Grennan <tmgrennan@gmail.com>
Subject: Re: [PATCH 3/8] tag: Change misleading --list <pattern> documentation
Date: Mon, 20 Mar 2017 12:11:34 -0400	[thread overview]
Message-ID: <20170320161134.7g5ljysunwsjiz3l@sigill.intra.peff.net> (raw)
In-Reply-To: <CACBZZX5V3fMpwUCFr8vXMsxkXpz353-BOtaANYvnu9Qnz1GSQg@mail.gmail.com>

On Mon, Mar 20, 2017 at 05:09:02PM +0100, Ævar Arnfjörð Bjarmason wrote:

> > I think it's expected to work under the usual last-one-wins option
> > parsing. A more subtle case is that:
> >
> >   git tag -l -d foo
> >
> > would override "-l" with "-d". That's reasonable under the same rule as
> > long as the user knows that the two are mode-selectors. I don't think we
> > make that explicit in the documentation, though, so perhaps it isn't
> > something users should rely on.
> 
> That hasn't been the case since v1.8.5 (v1.8.4-rc0-12-g1158826394).
> Now supplying multiple CMDMODE invocations will die.
> 
> It is the case that we still need to manually check any pseudo-cmdmode
> switches like "tag -a" (bool) v.s. "tag -l" (cmdmode). We check that
> particular combination, but we doubtless have bugs like that in other
> commands.

Yeah, you're right. I didn't look past the options[] array to see that
we handled this specially. I guess there's nothing to document, then.

-Peff