git-merge-subordinate

git-merge-subordinate

[Matthew Wilcox]

Linus doesn't like seeing unnecessary merges in his tree.  I'm not a huge
fan of them either.  Wouldn't it be nice if we had a merge method that
did a merge without creating a merge?  I call it git-merge-subordinate
(since my tree is subordinate to the tree I'm pulling from).  I suppose
you could call it 'slave' if you want to be more pithy.  Anyway, this
is a first attempt, and it's totally cargo-cult programming; I make no
claim that I understand what I'm doing.  But it does seem to work.

While working on it, I found a small bug in git-merge.  When
no_trivial_merge_strategies has more than one component, setting
index_merge to f doesn't work.  So first, here's the patch to git-merge
adding support for 'subordinate':

--- /usr/bin/git-merge	2006-07-29 15:47:09.000000000 -0600
+++ /home/willy/bin/git-merge	2006-10-25 09:21:00.000000000 -0600
@@ -9,15 +9,15 @@
 LF='
 '
 
-all_strategies='recursive octopus resolve stupid ours'
+all_strategies='recursive octopus resolve stupid subordinate ours'
 default_twohead_strategies='recursive'
 default_octopus_strategies='octopus'
-no_trivial_merge_strategies='ours'
+no_trivial_merge_strategies='subordinate ours'
 use_strategies=
 
 index_merge=t
 if test ""; then
-	all_strategies='resolve octopus stupid ours'
+	all_strategies='resolve octopus stupid subordinate ours'
 	default_twohead_strategies='resolve'
 fi
 
@@ -154,12 +154,15 @@
 
 for s in $use_strategies
 do
-	case " $s " in
-	*" $no_trivial_merge_strategies "*)
-		index_merge=f
-		break
-		;;
-	esac
+	for t in $no_trivial_merge_strategies
+	do
+		case "$s" in
+		"$t")
+			index_merge=f
+			break
+			;;
+		esac
+	done
 done
 
 case "$#" in


And now, here's the extremely lame git-merge-subordinate script.

#!/bin/sh
#
# Copyright (c) 2005 Linus Torvalds
# Copyright (c) 2005 Junio C Hamano
#
# Resolve two trees, using enhancd multi-base read-tree.

# The first parameters up to -- are merge bases; the rest are heads.
bases= head= remotes= sep_seen=
for arg
do
	case ",$sep_seen,$head,$arg," in
	*,--,)
		sep_seen=yes
		;;
	,yes,,*)
		head=$arg
		;;
	,yes,*)
		remotes="$remotes$arg "
		;;
	*)
		bases="$bases$arg "
		;;
	esac
done

# Give up if we are given more than two remotes -- not handling octopus.
case "$remotes" in
?*' '?*)
	exit 2 ;;
esac

# Give up if this is a baseless merge.
if test '' = "$bases"
then
	exit 2
fi

git-rebase $remotes || exit 2
if result_tree=$(git-write-tree  2>/dev/null)
then
	exit 0
else
	echo "Simple merge failed, trying Automatic merge."
	if git-merge-index -o git-merge-one-file -a
	then
		exit 0
	else
		exit 1
	fi
fi

Re: git-merge-subordinate

[Jakub Narebski]

Matthew Wilcox wrote:

> Linus doesn't like seeing unnecessary merges in his tree.  I'm not a huge
> fan of them either.  Wouldn't it be nice if we had a merge method that
> did a merge without creating a merge?  I call it git-merge-subordinate
> (since my tree is subordinate to the tree I'm pulling from).  I suppose
> you could call it 'slave' if you want to be more pithy.  Anyway, this
> is a first attempt, and it's totally cargo-cult programming; I make no
> claim that I understand what I'm doing.  But it does seem to work.

Hmmm... the --squash option to git-merge/git-pull isn't enough?

--squash::
        Produce the working tree and index state as if a real
        merge happened, but do not actually make a commit or
        move the `HEAD`, nor record `$GIT_DIR/MERGE_HEAD` to
        cause the next `git commit` command to create a merge
        commit.  This allows you to create a single commit on
        top of the current branch whose effect is the same as
        merging another branch (or more in case of an octopus).


-- 
Jakub Narebski
Warsaw, Poland
ShadeHawk on #git

Re: git-merge-subordinate

[Johannes Schindelin]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 799 bytes --]

Hi,

On Wed, 25 Oct 2006, Jakub Narebski wrote:

> Matthew Wilcox wrote:
> 
> > Linus doesn't like seeing unnecessary merges in his tree.  I'm not a huge
> > fan of them either.  Wouldn't it be nice if we had a merge method that
> > did a merge without creating a merge?  I call it git-merge-subordinate
> > (since my tree is subordinate to the tree I'm pulling from).  I suppose
> > you could call it 'slave' if you want to be more pithy.  Anyway, this
> > is a first attempt, and it's totally cargo-cult programming; I make no
> > claim that I understand what I'm doing.  But it does seem to work.
> 
> Hmmm... the --squash option to git-merge/git-pull isn't enough?

What subordinate does is not _merge_, but _rebase_ on top of the fetched 
commit. So yes, --squash isn't enough ;-)

Ciao,
Dscho

Re: git-merge-subordinate

[Shawn Pearce]

Johannes Schindelin <Johannes.Schindelin@gmx.de> wrote:
> Hi,
> 
> On Wed, 25 Oct 2006, Jakub Narebski wrote:
> 
> > Matthew Wilcox wrote:
> > 
> > > Linus doesn't like seeing unnecessary merges in his tree.  I'm not a huge
> > > fan of them either.  Wouldn't it be nice if we had a merge method that
> > > did a merge without creating a merge?  I call it git-merge-subordinate
> > > (since my tree is subordinate to the tree I'm pulling from).  I suppose
> > > you could call it 'slave' if you want to be more pithy.  Anyway, this
> > > is a first attempt, and it's totally cargo-cult programming; I make no
> > > claim that I understand what I'm doing.  But it does seem to work.
> > 
> > Hmmm... the --squash option to git-merge/git-pull isn't enough?
> 
> What subordinate does is not _merge_, but _rebase_ on top of the fetched 
> commit. So yes, --squash isn't enough ;-)


And I would suggest calling it 'git-merge-rebase', as the strategy
really is rebase...  :-)

--