[PATCH v2 0/2] Propagating flags carefully from the command line

[PATCH v2 0/2] Propagating flags carefully from the command line

[Junio C Hamano]

So this is my second try.  The second one now gets rid of the call
to mark_blob_uninteresting() as Peff suggested, because the first
patch makes the function very well aware that it only should mark
the objects that are reachable from the object, and by definition
blobs do not reach anything.

Junio C Hamano (2):
  revision: mark contents of an uninteresting tree uninteresting
  revision: propagate flag bits from tags to pointees

 revision.c               | 29 +++++++++++++++++------------
 t/t6000-rev-list-misc.sh | 17 +++++++++++++++++
 2 files changed, 34 insertions(+), 12 deletions(-)

-- 
1.8.5.3-493-gb139ac2

[PATCH v2 1/2] revision: mark contents of an uninteresting tree uninteresting

[Junio C Hamano]

"git rev-list --objects ^A^{tree} B^{tree}" ought to mean "I want a
list of objects inside B's tree, but please exclude the objects that
appear inside A's tree".

we see the top-level tree marked as uninteresting (i.e. ^A^{tree} in
the above example) and call mark_tree_uninteresting() on it; this
unfortunately prevents us from recursing into the tree and marking
the objects in the tree as uninteresting.

The reason why "git log ^A A" yields an empty set of commits,
i.e. we do not have a similar issue for commits, is because we call
mark_parents_uninteresting() after seeing an uninteresting commit.
The uninteresting-ness of the commit itself does not prevent its
parents from being marked as uninteresting.

Introduce mark_tree_contents_uninteresting() and structure the code
in handle_commit() in such a way that it makes it the responsibility
of the callchain leading to this function to mark commits, trees and
blobs as uninteresting, and also make it the responsibility of the
helpers called from this function to mark objects that are reachable
from them.

Note that this is a very old bug that probably dates back to the day
when "rev-list --objects" was introduced.  The line to clear
tree->object.parsed at the end of mark_tree_contents_uninteresting()
can be removed when this fix is merged to the codebase after
6e454b9a (clear parsed flag when we free tree buffers, 2013-06-05).

Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 revision.c               | 25 +++++++++++++++++--------
 t/t6000-rev-list-misc.sh |  6 ++++++
 2 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/revision.c b/revision.c
index 7010aff..28449c5 100644
--- a/revision.c
+++ b/revision.c
@@ -98,17 +98,12 @@ static void mark_blob_uninteresting(struct blob *blob)
 	blob->object.flags |= UNINTERESTING;
 }
 
-void mark_tree_uninteresting(struct tree *tree)
+static void mark_tree_contents_uninteresting(struct tree *tree)
 {
 	struct tree_desc desc;
 	struct name_entry entry;
 	struct object *obj = &tree->object;
 
-	if (!tree)
-		return;
-	if (obj->flags & UNINTERESTING)
-		return;
-	obj->flags |= UNINTERESTING;
 	if (!has_sha1_file(obj->sha1))
 		return;
 	if (parse_tree(tree) < 0)
@@ -135,6 +130,19 @@ void mark_tree_uninteresting(struct tree *tree)
 	 */
 	free(tree->buffer);
 	tree->buffer = NULL;
+	tree->object.parsed = 0;
+}
+
+void mark_tree_uninteresting(struct tree *tree)
+{
+	struct object *obj = &tree->object;
+
+	if (!tree)
+		return;
+	if (obj->flags & UNINTERESTING)
+		return;
+	obj->flags |= UNINTERESTING;
+	mark_tree_contents_uninteresting(tree);
 }
 
 void mark_parents_uninteresting(struct commit *commit)
@@ -294,7 +302,8 @@ static struct commit *handle_commit(struct rev_info *revs, struct object *object
 		if (!revs->tree_objects)
 			return NULL;
 		if (flags & UNINTERESTING) {
-			mark_tree_uninteresting(tree);
+			tree->object.flags |= UNINTERESTING;
+			mark_tree_contents_uninteresting(tree);
 			return NULL;
 		}
 		add_pending_object(revs, object, "");
@@ -309,7 +318,7 @@ static struct commit *handle_commit(struct rev_info *revs, struct object *object
 		if (!revs->blob_objects)
 			return NULL;
 		if (flags & UNINTERESTING) {
-			mark_blob_uninteresting(blob);
+			blob->object.flags |= UNINTERESTING;
 			return NULL;
 		}
 		add_pending_object(revs, object, "");
diff --git a/t/t6000-rev-list-misc.sh b/t/t6000-rev-list-misc.sh
index 15e3d64..9ad4971 100755
--- a/t/t6000-rev-list-misc.sh
+++ b/t/t6000-rev-list-misc.sh
@@ -56,4 +56,10 @@ test_expect_success 'rev-list A..B and rev-list ^A B are the same' '
 	test_cmp expect actual
 '
 
+test_expect_success 'propagate uninteresting flag down correctly' '
+	git rev-list --objects ^HEAD^{tree} HEAD^{tree} >actual &&
+	>expect &&
+	test_cmp expect actual
+'
+
 test_done
-- 
1.8.5.3-493-gb139ac2

[PATCH v2 2/2] revision: propagate flag bits from tags to pointees

[Junio C Hamano]

With the previous fix 895c5ba3 (revision: do not peel tags used in
range notation, 2013-09-19), handle_revision_arg() that processes
command line arguments for the "git log" family of commands no
longer directly places the object pointed by the tag in the pending
object array when it sees a tag object.  We used to place pointee
there after copying the flag bits like UNINTERESTING and
SYMMETRIC_LEFT.

This change meant that any flag that is relevant to later history
traversal must now be propagated to the pointed objects (most often
these are commits) while starting the traversal, which is partly
done by handle_commit() that is called from prepare_revision_walk().
We did propagate UNINTERESTING, but did not do so for others, most
notably SYMMETRIC_LEFT.  This caused "git log --left-right v1.0..."
(where "v1.0" is a tag) to start losing the "leftness" from the
commit the tag points at.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 revision.c               |  8 ++------
 t/t6000-rev-list-misc.sh | 11 +++++++++++
 2 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/revision.c b/revision.c
index 28449c5..aec0333 100644
--- a/revision.c
+++ b/revision.c
@@ -273,6 +273,7 @@ static struct commit *handle_commit(struct rev_info *revs, struct object *object
 				return NULL;
 			die("bad object %s", sha1_to_hex(tag->tagged->sha1));
 		}
+		object->flags |= flags;
 	}
 
 	/*
@@ -284,7 +285,6 @@ static struct commit *handle_commit(struct rev_info *revs, struct object *object
 		if (parse_commit(commit) < 0)
 			die("unable to parse commit %s", name);
 		if (flags & UNINTERESTING) {
-			commit->object.flags |= UNINTERESTING;
 			mark_parents_uninteresting(commit);
 			revs->limited = 1;
 		}
@@ -302,7 +302,6 @@ static struct commit *handle_commit(struct rev_info *revs, struct object *object
 		if (!revs->tree_objects)
 			return NULL;
 		if (flags & UNINTERESTING) {
-			tree->object.flags |= UNINTERESTING;
 			mark_tree_contents_uninteresting(tree);
 			return NULL;
 		}
@@ -314,13 +313,10 @@ static struct commit *handle_commit(struct rev_info *revs, struct object *object
 	 * Blob object? You know the drill by now..
 	 */
 	if (object->type == OBJ_BLOB) {
-		struct blob *blob = (struct blob *)object;
 		if (!revs->blob_objects)
 			return NULL;
-		if (flags & UNINTERESTING) {
-			blob->object.flags |= UNINTERESTING;
+		if (flags & UNINTERESTING)
 			return NULL;
-		}
 		add_pending_object(revs, object, "");
 		return NULL;
 	}
diff --git a/t/t6000-rev-list-misc.sh b/t/t6000-rev-list-misc.sh
index 9ad4971..3794e4c 100755
--- a/t/t6000-rev-list-misc.sh
+++ b/t/t6000-rev-list-misc.sh
@@ -62,4 +62,15 @@ test_expect_success 'propagate uninteresting flag down correctly' '
 	test_cmp expect actual
 '
 
+test_expect_success 'symleft flag bit is propagated down from tag' '
+	git log --format="%m %s" --left-right v1.0...master >actual &&
+	cat >expect <<-\EOF &&
+	> two
+	> one
+	< another
+	< that
+	EOF
+	test_cmp expect actual
+'
+
 test_done
-- 
1.8.5.3-493-gb139ac2

Re: [PATCH v2 1/2] revision: mark contents of an uninteresting tree uninteresting

[Jonathan Nieder]

Junio C Hamano wrote:

> we see the top-level tree marked as uninteresting (i.e. ^A^{tree} in
> the above example) and call mark_tree_uninteresting() on it; this
> unfortunately prevents us from recursing into the tree and marking
> the objects in the tree as uninteresting.

So the tree is marked uninteresting twice --- once by setting in the
UNINTERESTING flag in handle_revision_arg() and a second attempted
time in mark_tree_uninteresting()?   Makes sense.

Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>

Re: [PATCH v2 1/2] revision: mark contents of an uninteresting tree uninteresting

[Junio C Hamano]

Jonathan Nieder <jrnieder@gmail.com> writes:

> Junio C Hamano wrote:
>
>> we see the top-level tree marked as uninteresting (i.e. ^A^{tree} in
>> the above example) and call mark_tree_uninteresting() on it; this
>> unfortunately prevents us from recursing into the tree and marking
>> the objects in the tree as uninteresting.
>
> So the tree is marked uninteresting twice --- once by setting in the
> UNINTERESTING flag in handle_revision_arg() and a second attempted
> time in mark_tree_uninteresting()?   Makes sense.

It is that the original code, the setting of the mark on the object
itself was inconsistent.  For commits, we did that ourselves; for
trees, we let the mark_tree_uninteresting() do so.

And mark_tree_uninteresting() has this quirk that it refuses to
recurse into the given tree, if the tree is already marked as
uninteresting by the caller.

We did not have the same problem on commits, because we make a
similar call to mark-parents-uninteresting but the function does not
care if the commit itself is already marked as uninteresting.

The distinction does not matter when tags are not involved.  But
once we start propagating the flags from a tag to objects that the
tag points at, it starts to matter.  The caller will mark the object
uninteresting in the loop that peels the tag, and the resulting
object is uninteresting.  It is not a problem for commits.  It was a
problem for trees, which used mark_tree_uninteresting() to mark all
the objects inside the tree uninteresting, including the tree itself.

Re: [PATCH v2 0/2] Propagating flags carefully from the command line

[Jeff King]

On Wed, Jan 15, 2014 at 03:59:42PM -0800, Junio C Hamano wrote:

> So this is my second try.  The second one now gets rid of the call
> to mark_blob_uninteresting() as Peff suggested, because the first
> patch makes the function very well aware that it only should mark
> the objects that are reachable from the object, and by definition
> blobs do not reach anything.
> 
> Junio C Hamano (2):
>   revision: mark contents of an uninteresting tree uninteresting
>   revision: propagate flag bits from tags to pointees

Sorry for a slow review, but I just read through your earlier comments
and this series. I agree there was definitely a bug in what we were
discussing earlier, and this looks like the right way to fix it. The end
result splits the flag-setting responsibility much more sensibly.

Thanks.

-Peff