From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Wed, 09 Mar 2022 22:20:30 GMT
Subject: Re: [PATCH] checkout, clone: die if tree cannot be parsed
Message-ID: <nycvar.QRO.7.76.6.2203092319060.357@tvgsbejvaqbjf.bet>
In-Reply-To: <kl6lilswdu68.fsf@chooglen-macbookpro.roam.corp.google.com>

Hi Glen,

On Wed, 2 Mar 2022, Glen Choo wrote:

> Junio C Hamano <gitster@pobox.com> writes:
>
> >          but one random-looking hexadecimal string is almost
> > indistinguishable from another, and neither would be a very useful
> > message from the end user's point of view.  I am wondering if we can
> > use old_branch_info to formulate something easier to understand for
> > the user.  update_refs_for_switch() seems to compute old_desc as a
> > human readable name by using old_branch_info->name if available
> > before falling back to old_branch_info->commit object name, which
> > might be a source of inspiration.
>
> I think it's actually more helpful to have the oid instead of a
> human-readable description like old_branch_info->name.

The most helpful would be to have both. That way, it would at least be
potentially possible to figure out from a ref how to fetch a non-corrupt
version from elsewhere.

Ciao,
Dscho