Re: Possible bug with git restore

From: Junio C Hamano <gitster@pobox.com>
Date: Aug 22, 2020, 19:36 UTC
Message-ID: <xmqq364e79sm.fsf@gitster.c.googlers.com>
In-Reply-To: <20200820182720.GA2537643@coredump.intra.peff.net>

Jeff King <peff@peff.net> writes:

Show 14 quoted lines

> So I think this inconsistency in pathspec matching between trees and the
> index has probably existed in git-checkout for ages (and I guess people
> don't do wildcards with trees often enough for anybody to have noticed).
> But it didn't cause the index-deletion problem, because that only
> appeared more recently with the --no-overlay mode. That's the default
> for restore, but you can trigger the problem with checkout, too:
>
>   $ git reset --hard
>   $ git checkout --no-overlay HEAD^ '*.hpp'
>   Updated 0 paths from 2668463
>   $ git status
>   On branch master
>   Changes to be committed:
> 	deleted:    incl/test_file.hpp

The --no-overlay mode is an enhancement added on top of reasonably aged codebase relatively recently. Most of the core code in checkout dates back to early 2008, while --no-overlay was done as a long-overdue-afterthought in early last year.

And it is not all that surprising that this issue took a long time to be discovered.

Thank you all for finding, analysing and fixing it promptly.