finding similar blobs (was: Re: $Revision$ keyword replacement?

From: Junio C Hamano <junkio@cox.net>
Date: Nov 22, 2005, 22:42 UTC
Message-ID: <7vy83gb8zk.fsf@assigned-by-dhcp.cox.net>
In-Reply-To: <871x18h9ee.fsf@ifae.es>

Santi Béjar <sbejar@gmail.com> writes:

>         How similar this file is to the file $path in these commit(s)?

Very cute.

> tmp=`mktemp -t -d git-find-sim.XXXXXXX`
> ...
> git update-index --add $file || exit 1
> tree=`git-write-tree`

Are you going through all this trouble just to avoid a blob and a tree object left dangling after you are done? Or is there something else going on?

Show 6 quoted lines

> rev_arg=`GIT_DIR=$GIT_DIR_ORIG git-rev-parse --default HEAD --revs-only "$@"`
> revs=`GIT_DIR=$GIT_DIR_ORIG git-rev-list $rev_arg`
> for i in $revs; do
>     git diff-tree --name-status $i -C $tree | grep $file |
>     sed "s/^/$i:/"
> done

Perhaps

        GIT_DIR=$GIT_DIR_ORIG git-rev-list $rev_arg |
        while read one
            git diff-tree --name-status -r $one -C $tree | grep $file |
            sed "s/^/$one:/"
        done

just in case the similar file you will discover is hidden in a subdirectory?