From: Jeff King <peff@peff.net>
Date: Tue, 04 Feb 2020 19:54:01 GMT
Subject: Re: [PATCH v2 5/5] Reftable support for git-core
Message-ID: <20200204195401.GB12705@coredump.intra.peff.net>
In-Reply-To: <CAFQ2z_O3DhLbVtpGBWERnwGawtvE-iZNKRDGg_sfMhOX6g=Wpw@mail.gmail.com>

On Tue, Feb 04, 2020 at 08:06:57PM +0100, Han-Wen Nienhuys wrote:

> On Wed, Jan 29, 2020 at 11:47 AM Jeff King <peff@peff.net> wrote:
> 
> > That said, it might make for easier debugging if the reftables file
> > declares the size it assumes.
> 
> If there is a mismatch in size, the reftable will look completely
> corrupted data.  I think this will be a bad experience.

Yes, but I think it would take a bunch of other failures to get there.
And it's true of all of the other parts of Git, too; the master switch
is a config setting that says "I'm a sha-256 repository".

That said, I'm not at all opposed to a header in the reftables file
saying "I'm a sha-256 reftable". We shouldn't ever see a mismatch there,
but that's what I meant by "easier debugging".

-Peff