[PATCH 4/4] git-fetch-pack: Implement client part of the multi_ack extension

This patch concludes the series, which makes 
git-fetch-pack/git-upload-pack negotiate a potentially better set of 
common revs. It should make a difference when fetching from a repository 
with a few branches.

Signed-off-by: Johannes Schindelin <redacted>

---

 connect.c    |    5 ++++-
 fetch-pack.c |   50 +++++++++++++++++++++++++++++++++++++-------------
 2 files changed, 41 insertions(+), 14 deletions(-)

applies-to: 6b4b7d9acf60aa99d961b599f37d0c824be79e27
9adb6b3971e7daa79221d7dbe05b66327b266b86

diff --git a/connect.c b/connect.c
index b171c5d..57e25a3 100644
--- a/connect.c
+++ b/connect.c

@@ -59,8 +59,11 @@ int get_ack(int fd, unsigned char *resul
 	if (!strcmp(line, "NAK"))
 		return 0;
 	if (!strncmp(line, "ACK ", 3)) {
-		if (!get_sha1_hex(line+4, result_sha1))
+		if (!get_sha1_hex(line+4, result_sha1)) {
+			if (strstr(line+45, "continue"))
+				return 2;
 			return 1;
+		}
 	}
 	die("git-fetch_pack: expected ACK/NAK, got '%s'", line);
 }

diff --git a/fetch-pack.c b/fetch-pack.c
index 3a903c4..57602b9 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c

@@ -125,7 +125,7 @@ static int find_common(int fd[2], unsign
 		       struct ref *refs)
 {
 	int fetching;
-	int count = 0, flushes = 0, retval;
+	int count = 0, flushes = 0, multi_ack = 0, retval;
 	const unsigned char *sha1;
 
 	for_each_ref(rev_list_append_sha1);

@@ -156,20 +156,22 @@ static int find_common(int fd[2], unsign
 			continue;
 		}
 
-		packet_write(fd[1], "want %s\n", sha1_to_hex(remote));
+		packet_write(fd[1], "want %s multi_ack\n", sha1_to_hex(remote));
 		fetching++;
 	}
 	packet_flush(fd[1]);
 	if (!fetching)
 		return 1;
 
-	flushes = 1;
+	flushes = 0;
 	retval = -1;
 	while ((sha1 = get_rev())) {
 		packet_write(fd[1], "have %s\n", sha1_to_hex(sha1));
 		if (verbose)
 			fprintf(stderr, "have %s\n", sha1_to_hex(sha1));
 		if (!(31 & ++count)) {
+			int ack;
+
 			packet_flush(fd[1]);
 			flushes++;
 

@@ -179,26 +181,48 @@ static int find_common(int fd[2], unsign
 			 */
 			if (count == 32)
 				continue;
-			if (get_ack(fd[0], result_sha1)) {
-				flushes = 0;
-				retval = 0;
-				if (verbose)
-					fprintf(stderr, "got ack\n");
-				break;
-			}
+
+			do {
+				ack = get_ack(fd[0], result_sha1);
+				if (verbose && ack)
+					fprintf(stderr, "got ack %d %s\n", ack,
+							sha1_to_hex(result_sha1));
+				if (ack == 1) {
+					if (!multi_ack)
+						flushes = 0;
+					retval = 0;
+					goto done;
+				} else if (ack == 2) {
+					multi_ack = 1;
+					mark_common((struct commit *)
+							lookup_object(result_sha1));
+					retval = 0;
+				}
+			} while(ack);
 			flushes--;
 		}
 	}
+done:
+	if (multi_ack) {
+		packet_flush(fd[1]);
+		flushes++;
+	}
 	packet_write(fd[1], "done\n");
 	if (verbose)
 		fprintf(stderr, "done\n");
+	if (retval != 0)
+		flushes++;
 	while (flushes) {
-		flushes--;
 		if (get_ack(fd[0], result_sha1)) {
 			if (verbose)
-				fprintf(stderr, "got ack\n");
-			return 0;
+				fprintf(stderr, "got ack %s\n",
+					sha1_to_hex(result_sha1));
+			if (!multi_ack)
+				return 0;
+			retval = 0;
+			continue;
 		}
+		flushes--;
 	}
 	return retval;
 }

---
0.99.8.GIT

Johannes Schindelin, Sun, Oct 23, 2005 03:40:13 +0200:

This patch concludes the series, which makes 
git-fetch-pack/git-upload-pack negotiate a potentially better set of 
common revs. It should make a difference when fetching from a repository 
with a few branches.

This broke git-pull for me (the local one):

    /d/e/f.git$ git-pull
    fatal: bad pack file
    fatal: git-unpack-objects died with error code 128
    Fetch failure: /a/b/c/.git

applies-to: 6b4b7d9acf60aa99d961b599f37d0c824be79e27
9adb6b3971e7daa79221d7dbe05b66327b266b86

...

Reverting just fetch-pack.c part of the patch helps.

Hi,

On Tue, 25 Oct 2005, Alex Riesen wrote:

Johannes Schindelin, Sun, Oct 23, 2005 03:40:13 +0200:

quoted

This patch concludes the series, which makes 
git-fetch-pack/git-upload-pack negotiate a potentially better set of 
common revs. It should make a difference when fetching from a repository 
with a few branches.

This broke git-pull for me (the local one):

    /d/e/f.git$ git-pull
    fatal: bad pack file
    fatal: git-unpack-objects died with error code 128
    Fetch failure: /a/b/c/.git

quoted

applies-to: 6b4b7d9acf60aa99d961b599f37d0c824be79e27
9adb6b3971e7daa79221d7dbe05b66327b266b86

...

quoted

diff --git a/fetch-pack.c b/fetch-pack.c
index 3a903c4..57602b9 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c

Reverting just fetch-pack.c part of the patch helps.

Could you please try the patch I sent with the subject "[PATCH] 
fetch/upload: Fix corner case with few revs"? Your output looks exactly 
like what I fixed with that patch.

Ciao,
Dscho

On 10/25/05, Johannes Schindelin [off-list ref] wrote:

quoted

quoted

This patch concludes the series, which makes
git-fetch-pack/git-upload-pack negotiate a potentially better set of
common revs. It should make a difference when fetching from a repository
with a few branches.

This broke git-pull for me (the local one):

    /d/e/f.git$ git-pull
    fatal: bad pack file
    fatal: git-unpack-objects died with error code 128
    Fetch failure: /a/b/c/.git

quoted

applies-to: 6b4b7d9acf60aa99d961b599f37d0c824be79e27
9adb6b3971e7daa79221d7dbe05b66327b266b86

...

quoted

diff --git a/fetch-pack.c b/fetch-pack.c
index 3a903c4..57602b9 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c

Reverting just fetch-pack.c part of the patch helps.

Could you please try the patch I sent with the subject "[PATCH]
fetch/upload: Fix corner case with few revs"? Your output looks exactly
like what I fixed with that patch.

I couldn't at the moment. Do you still need a test?

Hi,

On Wed, 26 Oct 2005, Alex Riesen wrote:

quoted

Could you please try the patch I sent with the subject "[PATCH]
fetch/upload: Fix corner case with few revs"? Your output looks exactly
like what I fixed with that patch.

I couldn't at the moment. Do you still need a test?

If you have time and can test it, yes, please.

Ciao,
Dscho

On 10/26/05, Johannes Schindelin [off-list ref] wrote:

quoted

quoted

Could you please try the patch I sent with the subject "[PATCH]
fetch/upload: Fix corner case with few revs"? Your output looks exactly
like what I fixed with that patch.

I couldn't at the moment. Do you still need a test?

If you have time and can test it, yes, please.

Will try. Which patch is it?