Re: git reset --hard and uncommitted files in index

From: Johannes Sixt <j.sixt@viscovery.net>
Date: May 22, 2009, 06:43 UTC
Message-ID: <4A1649A9.5090606@viscovery.net>
In-Reply-To: <gv52s4$e7h$1@ger.gmane.org>

Thomas Guyot-Sionnest schrieb:

Show 13 quoted lines

> Today I was working on a script that modify many files on a repository,
> therefore during testing phase I needed to reset --hard to get back to
> the last commit. This operation does not affect untracked files.
> 
> Since i was going to commit the script, at some point I decided to add
> it to the index so I could track/revert changes to it. On the next reset
> Git deleted the script.
> 
> Fortunately "git add" creates an object that I could easily recover
> (thanks to wereHamster on #git for pointing that out), however I was
> wondering if it's really the desired behavior to *delete* a file that
> have been added to the index but isn't on the previous commit? Shouldn't
> git just unstage it?

No, the behavior is by design. Its purpose is to return the working directory to a clean slate. For example, after a conflicted merge the merged branch may have brought new files, but you wouldn't want to keep them around after the cleanup.

'git reset --hard' is "do what I say", not "do what I mean". ;-)

-- Hannes