On Fri, Sep 10, 2010 at 04:54:11PM +0200, Steven wrote: > > You could just manually do the revert. Something like: > > > > git diff-tree -p $commit | git apply --ignore-whitespace > > git commit -m "revert '`git log -1 --format=%s $commit`'" > > I had to modify the commands a bit to get it to work. > Here they are: > git diff-tree -p <commithash> | git apply --reverse --ignore-whitespace -C0 > git add <file(s)> > git commit -m "revert '`git log -1 --format=%s $commit`'" > > The --reverse is necessary to revert a patch, I needed the -C0 parameter > as well because the line above changed as well. Oops. Yeah, obviously I just typed that straight into the email and did not actually run it. :) The --reverse is definitely necessary. Using -C0 can help, but it can also be dangerous, as context lines help apply make sure it's in the right spot. > This was a fairly simple example, but I imagine it won't work at all with > a larger history, especially with more changes in the relevant sections > and additions/deletions. I believe git revert does take these into > account? Yeah, the question is really whether the reverse diff from the reverted commit applies to your current tree. Nearby changes obviously make that harder. Once upon a time revert itself was implemented like this (see 045f82c, which introduced revert and uses "diff | apply"). These days it is only a little more complex. It uses the 3-way merge machinery, which should do better with finding minimal conflicts when the context has changed, and will do rename detection (and when the patch doesn't apply, will actually put in conflict markers, which is a nice place to start with resolving it). Revert is written in C these days, but you can see the shell script version using git-merge-recursive in contrib/examples/git-revert.sh. However, I don't think there is an easy way to ask merge-recursive to ignore whitespace changes. -Peff -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html