Re: [PATCH v6 6/6] convert: add "status=delayed" to filter process protocol

From: Lars Schneider <larsxschneider@gmail.com>
To: Junio C Hamano <gitster@pobox.com>
Cc: "Git Users" <git@vger.kernel.org>, "Jeff King" <peff@peff.net>,
	"Torsten Bögershausen" <tboegi@web.de>, "Eric Wong" <e@80x24.org>,
	"Taylor Blau" <ttaylorr@github.com>,
	peartben@gmail.com
Subject: Re: [PATCH v6 6/6] convert: add "status=delayed" to filter process protocol
Date: Tue, 27 Jun 2017 14:07:01 +0200	[thread overview]
Message-ID: <22F3047C-F2B2-4327-97B4-3C2347E1DD0E@gmail.com> (raw)
In-Reply-To: <xmqqa84u4cuk.fsf@gitster.mtv.corp.google.com>

> On 27 Jun 2017, at 00:13, Junio C Hamano <gitster@pobox.com> wrote:
> 
> Lars Schneider <larsxschneider@gmail.com> writes:
> 
>> ...
> 
>>> I am wondering whose responsibility it will be to deal with a path
>>> "list-available" reports that are *not* asked by Git or Git got no
>>> "delayed" response.  The list subtraction done by the caller is
>>> probably the logical place to do so.
>> 
>> Correct. Git (the caller) will notice that not all "delayed" paths
>> are listed by the filter and throw an error at the end.
> 
> I am wondering about the other case.  Git didn't ask for a path to
> be filtered at all, but the filter sneaked in a path that happens to
> in the index in its response---Git should at least ignore it, and
> better yet, diagnose it as an error in the filter process.

Agreed. I've used your implementation suggestion [1] and added two
test cases to ensure we signal a proper error in case of a buggy filter.

Thanks,
Lars

[1] http://public-inbox.org/git/xmqqzicu2x4a.fsf@gitster.mtv.corp.google.com/