Re: [PATCH] Make git revert warn the user when reverting a merge commit.

lördag 20 december 2008 23:54:19 skrev Boyd Stephen Smith Jr.:

On Saturday 2008 December 20 01:08:01 Robin Rosenberg wrote:

quoted

fredag 19 december 2008 03:39:15 skrev Boyd Stephen Smith Jr.:

quoted

On Thursday 2008 December 18 18:21:25 Linus Torvalds wrote:

quoted

I suspect we should warn about reverting merges.

Or mention the reverted parent in the commit message since it is not
obvious.

---
 builtin-revert.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/builtin-revert.c b/builtin-revert.c
index 4038b41..fc59229 100644
--- a/builtin-revert.c
+++ b/builtin-revert.c

@@ -352,6 +352,10 @@ static int revert_or_cherry_pick(int argc, const char

**argv) add_to_msg(oneline_body + 1);
 		add_to_msg("\"\n\nThis reverts commit ");
 		add_to_msg(sha1_to_hex(commit->object.sha1));
+		if (commit->parents->next) {
+			add_to_msg(" removing\ncontributions from ");
+			add_to_msg(sha1_to_hex(parent->object.sha1));
+		}
 		add_to_msg(".\n");
 	} else {
 		base = parent;

I'm still new to the code, but parent is the "mainline" specified on the 
command-line, which (I think) is actually the parent to be reverted to, so we 
are actually removing contributions from all the *other* parents.  So, the 
message may be backward.  Because of that, I'd say the patch doesn't handle 

Indeed the message is backward. How about  "removing all contributions except from"... etc ?

An alternative, would be "removing changes relative to .." (mainline). The changes are
the contributions from all other parents. I have to huge interest in the exact phrase used.

octopus merges well, either.

Same problem, I think.

-- robin