Re: What's cooking in git.git (Jul 2016, #02; Wed, 6)

From: Charles Bailey <charles@hashpling.org>
Date: Jul 7, 2016, 20:40 UTC
Message-ID: <20160707204002.GA19138@hashpling.org>
In-Reply-To: <xmqqr3b6qwdt.fsf@gitster.mtv.corp.google.com>

On Wed, Jul 06, 2016 at 02:39:26PM -0700, Junio C Hamano wrote:

Show 12 quoted lines

> * nd/ita-cleanup (2016-07-01) 3 commits
>   (merged to 'next' on 2016-07-06 at f15aeba)
>  + grep: fix grepping for "intent to add" files
>  + t7810-grep.sh: fix a whitespace inconsistency
>  + t7810-grep.sh: fix duplicated test name
> 
>  Git does not know what the contents in the index should be for a
>  path added with "git add -N" yet, so "git grep --cached" should not
>  show hits (or show lack of hits, with -L) in such a path.  But we
>  did by mistake, which has been corrected.
> 
>  Will merge to 'master'.

I just wanted to clarify what was actually fixed. The actual bug that was reported and fixed was the fact that 'git grep' (without --cached) wasn't searching the contents of files in the working tree if the index entry had the "intent to add" bit set.

My original proposed fix (reversion of the commit that introduced this change) caused the re-introduction of behaviour where "i-to-a" files would be reported with -L --cached which wasn't desired. I think because we spent most of our energy and discussion on fixing this in a better way we've lost the intent of the original patch.