git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
From: Junio C Hamano <gitster@pobox.com>
To: Miklos Vajna <vmiklos@frugalware.org>
Cc: git@vger.kernel.org
Subject: Re: [PATCH 09/10] Introduce get_octopus_merge_bases() in commit.c
Date: Thu, 05 Jun 2008 22:53:59 -0700	[thread overview]
Message-ID: <7vmylzw1l4.fsf@gitster.siamese.dyndns.org> (raw)
In-Reply-To: <7vk5h3xlt9.fsf@gitster.siamese.dyndns.org> (Junio C. Hamano's message of "Thu, 05 Jun 2008 20:51:46 -0700")

Junio C Hamano <gitster@pobox.com> writes:

> When you want to merge A, B and C, because internally we always do
> pairwise, you would first compute:
>
> 	T(A,B) == merge_3way(mb(A,B), A, B)
>
> and then you would want to pick some merge base to merge that result with
> C:
>
> 	T(A,B,C) == merge_3way(???, T(A,B), C)
>
> I do not think using mb(mb(A,B),C) as ??? is necessary nor optimal.
>
>           o---o---A
>          /         .
>     ----1---2---B...T(A,B)
>              \       .
>               o---C...T(A,B,C)
>
> "1" above is mb(A,B) (i.e. the merge base between A and B) that you would
> use when you merge A and B to compute an internal merge result T(A,B).
> And "1" also is mb(mb(A,B),C).
>
> But if you are doing 3-way merge to arrive at T(A,B,C) by merging T(A,B)
> and C, don't you want to be using "2" as the merge base, not "1"?
>
> The big comment on $MRC at the end of git-merge-octopus is about this
> issue.  In earlier implementation, we used to use "1" to create T(A,B) and
> then also used it to update the variable $MRC, which is used to compute
> the merge base to use when the tentative tree T(A,B) is merged with the
> other remote in the next round.  It was a mistake.
>
> Ideally we should pick a better base between mb(A,C) and mb(B,C) (and if
> we used mb(B,C) that's "2" and is a much better base than "1" when merging
> T(A,B) and C).  Using mb(mb(A,B),C) means we are guaranteed to pick the
> worst base for that last merge.

Something like this might be enough to get us started.  Instead of finding
out the merge bases between two commits, "one" and "two", this would find
merge base between "one" and a commit that would be a merge across all the
"two"s.  To apply it to the above example, you would give "C" as "one",
and "A" and "B" as "twos" to it, to compute "2".

 commit.c |   32 ++++++++++++++++++++++----------
 1 files changed, 22 insertions(+), 10 deletions(-)

diff --git a/commit.c b/commit.c
index 94d5b3d..816eed1 100644
--- a/commit.c
+++ b/commit.c
@@ -531,26 +531,33 @@ static struct commit *interesting(struct commit_list *list)
 	return NULL;
 }
 
-static struct commit_list *merge_bases(struct commit *one, struct commit *two)
+static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 {
 	struct commit_list *list = NULL;
 	struct commit_list *result = NULL;
+	int i;
 
-	if (one == two)
-		/* We do not mark this even with RESULT so we do not
-		 * have to clean it up.
-		 */
-		return commit_list_insert(one, &result);
+	for (i = 0; i < n; i++) {
+		if (one == twos[i])
+			/* We do not mark this even with RESULT so we do not
+			 * have to clean it up.
+			 */
+			return commit_list_insert(one, &result);
+	}
 
 	if (parse_commit(one))
 		return NULL;
-	if (parse_commit(two))
-		return NULL;
+	for (i = 0; i < n; i++) {
+		if (parse_commit(twos[i]))
+			return NULL;
+	}
 
 	one->object.flags |= PARENT1;
-	two->object.flags |= PARENT2;
 	insert_by_date(one, &list);
-	insert_by_date(two, &list);
+	for (i = 0; i < n; i++) {
+		twos[i]->object.flags |= PARENT2;
+		insert_by_date(twos[i], &list);
+	}
 
 	while (interesting(list)) {
 		struct commit *commit;
@@ -598,6 +605,11 @@ static struct commit_list *merge_bases(struct commit *one, struct commit *two)
 	return result;
 }
 
+static struct commit_list *merge_bases(struct commit *one, struct commit *two)
+{
+	return merge_bases_many(one, 1, &two);
+}
+
 struct commit_list *get_merge_bases(struct commit *one,
 					struct commit *two, int cleanup)
 {

  reply	other threads:[~2008-06-06  5:55 UTC|newest]

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-06-05 20:44 [PATCH 00/10] Build in merge Miklos Vajna
2008-06-05 20:44 ` [PATCH 01/10] Move split_cmdline() to alias.c Miklos Vajna
2008-06-05 20:44   ` [PATCH 02/10] Move commit_list_count() to commit.c Miklos Vajna
2008-06-05 20:44     ` [PATCH 03/10] Move builtin-remote's skip_prefix() to git-compat-util.h Miklos Vajna
2008-06-05 20:44       ` [PATCH 04/10] Add new test to ensure git-merge handles pull.twohead and pull.octopus Miklos Vajna
2008-06-05 20:44         ` [PATCH 05/10] parseopt: add a new PARSE_OPT_ARGV0_IS_AN_OPTION option Miklos Vajna
2008-06-05 20:44           ` [PATCH 06/10] Move read_cache_unmerged() to read-cache.c Miklos Vajna
2008-06-05 20:44             ` [PATCH 07/10] git-fmt-merge-msg: make it useable from other builtins Miklos Vajna
2008-06-05 20:44               ` [PATCH 08/10] Introduce commit_list_append() in commit.c Miklos Vajna
2008-06-05 20:44                 ` [PATCH 09/10] Introduce get_octopus_merge_bases() " Miklos Vajna
2008-06-05 20:44                   ` [PATCH 10/10] Build in merge Miklos Vajna
2008-06-06  3:51                   ` [PATCH 09/10] Introduce get_octopus_merge_bases() in commit.c Junio C Hamano
2008-06-06  5:53                     ` Junio C Hamano [this message]
2008-06-06 12:28                       ` Johannes Schindelin
2008-06-06 12:36                         ` Johannes Schindelin
2008-06-06 15:36                         ` Junio C Hamano
2008-06-07 21:38                       ` [PATCH] " Miklos Vajna
2008-06-09 14:02                         ` Johannes Schindelin
2008-06-09 22:43                           ` Miklos Vajna
2008-06-09 22:55                             ` Junio C Hamano
2008-06-09 23:08                               ` Johannes Schindelin
2008-06-09 23:20                                 ` Junio C Hamano
2008-06-09 23:35                                   ` Miklos Vajna
2008-06-09 23:06                             ` Junio C Hamano
2008-06-09 23:25                               ` Miklos Vajna
2008-06-09 23:31                               ` Johannes Schindelin
2008-06-09 23:41                                 ` Junio C Hamano
2008-06-10  0:03                                   ` Miklos Vajna
2008-06-10  2:43                                     ` Johannes Schindelin
2008-06-06 12:30                     ` [PATCH 09/10] " Johannes Schindelin
2008-06-07  2:30                     ` Miklos Vajna
2008-06-05 23:16                 ` [PATCH 08/10] Introduce commit_list_append() " Junio C Hamano
2008-06-06 23:52                   ` Miklos Vajna
2008-06-07  0:14                     ` Junio C Hamano
2008-06-07  2:03                       ` Miklos Vajna
2008-06-05 23:12               ` [PATCH 07/10] git-fmt-merge-msg: make it useable from other builtins Junio C Hamano
2008-06-07  1:04                 ` Miklos Vajna
2008-06-09  8:58               ` Andreas Ericsson
2008-06-09 22:53                 ` [PATCH] git-fmt-merge-msg: make it usable " Miklos Vajna
2008-06-05 23:05             ` [PATCH 06/10] Move read_cache_unmerged() to read-cache.c Junio C Hamano
2008-06-07  1:00               ` [PATCH] " Miklos Vajna
2008-06-05 22:58         ` [PATCH 04/10] Add new test to ensure git-merge handles pull.twohead and pull.octopus Junio C Hamano
2008-06-07  0:47           ` [PATCH] " Miklos Vajna
2008-06-05 22:38       ` [PATCH 03/10] Move builtin-remote's skip_prefix() to git-compat-util.h Junio C Hamano
2008-06-06 23:42         ` [PATCH] Move parse-options's " Miklos Vajna

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: http://vger.kernel.org/majordomo-info.html

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=7vmylzw1l4.fsf@gitster.siamese.dyndns.org \
    --to=gitster@pobox.com \
    --cc=git@vger.kernel.org \
    --cc=vmiklos@frugalware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://80x24.org/mirrors/git.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).