Smart fetch over HTTP: client side

The git-remote-curl backend detects if the remote server supports
the git-upload-pack service, and if so, runs git-fetch-pack locally
in a pipe to generate the want/have commands.

The advertisements from the server that were obtained during the
discovery are passed into git-fetch-pack before the POST request
starts, permitting server capability discovery and enablement.

Common objects that are discovered are appended onto the request as
have lines and are sent again on the next request.  This allows the
remote side to reinitialize its in-memory list of common objects
during the next request.

Because all requests are relatively short, below git-remote-curl's
1 MiB buffer limit, requests will use the standard Content-Length
header and be valid HTTP/1.0 POST requests.  This makes the fetch
client more tolerant of proxy servers which don't support HTTP/1.1
or the chunked transfer encoding.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
CC: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

1 files changed, 93 insertions, 17 deletions

diff --git a/builtin-fetch-pack.c b/builtin-fetch-pack.c
index 615f54974e..8ed4a6feaa 100644
--- a/builtin-fetch-pack.c
+++ b/builtin-fetch-pack.c
@@ -165,6 +165,24 @@ enum ack_type {
 	ACK_ready
 };
 
+static void consume_shallow_list(int fd)
+{
+	if (args.stateless_rpc && args.depth > 0) {
+		/* If we sent a depth we will get back "duplicate"
+		 * shallow and unshallow commands every time there
+		 * is a block of have lines exchanged.
+		 */
+		char line[1000];
+		while (packet_read_line(fd, line, sizeof(line))) {
+			if (!prefixcmp(line, "shallow "))
+				continue;
+			if (!prefixcmp(line, "unshallow "))
+				continue;
+			die("git fetch-pack: expected shallow list");
+		}
+	}
+}
+
 static enum ack_type get_ack(int fd, unsigned char *result_sha1)
 {
 	static char line[1000];
@@ -190,6 +208,15 @@ static enum ack_type get_ack(int fd, unsigned char *result_sha1)
 	die("git fetch_pack: expected ACK/NAK, got '%s'", line);
 }
 
+static void send_request(int fd, struct strbuf *buf)
+{
+	if (args.stateless_rpc) {
+		send_sideband(fd, -1, buf->buf, buf->len, LARGE_PACKET_MAX);
+		packet_flush(fd);
+	} else
+		safe_write(fd, buf->buf, buf->len);
+}
+
 static int find_common(int fd[2], unsigned char *result_sha1,
 		       struct ref *refs)
 {
@@ -199,7 +226,10 @@ static int find_common(int fd[2], unsigned char *result_sha1,
 	unsigned in_vain = 0;
 	int got_continue = 0;
 	struct strbuf req_buf = STRBUF_INIT;
+	size_t state_len = 0;
 
+	if (args.stateless_rpc && multi_ack == 1)
+		die("--stateless-rpc requires multi_ack_detailed");
 	if (marked)
 		for_each_ref(clear_marks, NULL);
 	marked = 1;
@@ -256,13 +286,13 @@ static int find_common(int fd[2], unsigned char *result_sha1,
 	if (args.depth > 0)
 		packet_buf_write(&req_buf, "deepen %d", args.depth);
 	packet_buf_flush(&req_buf);
-
-	safe_write(fd[1], req_buf.buf, req_buf.len);
+	state_len = req_buf.len;
 
 	if (args.depth > 0) {
 		char line[1024];
 		unsigned char sha1[20];
 
+		send_request(fd[1], &req_buf);
 		while (packet_read_line(fd[0], line, sizeof(line))) {
 			if (!prefixcmp(line, "shallow ")) {
 				if (get_sha1_hex(line + 8, sha1))
@@ -284,28 +314,40 @@ static int find_common(int fd[2], unsigned char *result_sha1,
 			}
 			die("expected shallow/unshallow, got %s", line);
 		}
+	} else if (!args.stateless_rpc)
+		send_request(fd[1], &req_buf);
+
+	if (!args.stateless_rpc) {
+		/* If we aren't using the stateless-rpc interface
+		 * we don't need to retain the headers.
+		 */
+		strbuf_setlen(&req_buf, 0);
+		state_len = 0;
 	}
 
 	flushes = 0;
 	retval = -1;
 	while ((sha1 = get_rev())) {
-		packet_write(fd[1], "have %s\n", sha1_to_hex(sha1));
+		packet_buf_write(&req_buf, "have %s\n", sha1_to_hex(sha1));
 		if (args.verbose)
 			fprintf(stderr, "have %s\n", sha1_to_hex(sha1));
 		in_vain++;
 		if (!(31 & ++count)) {
 			int ack;
 
-			packet_flush(fd[1]);
+			packet_buf_flush(&req_buf);
+			send_request(fd[1], &req_buf);
+			strbuf_setlen(&req_buf, state_len);
 			flushes++;
 
 			/*
 			 * We keep one window "ahead" of the other side, and
 			 * will wait for an ACK only on the next one
 			 */
-			if (count == 32)
+			if (!args.stateless_rpc && count == 32)
 				continue;
 
+			consume_shallow_list(fd[0]);
 			do {
 				ack = get_ack(fd[0], result_sha1);
 				if (args.verbose && ack)
@@ -322,6 +364,17 @@ static int find_common(int fd[2], unsigned char *result_sha1,
 				case ACK_continue: {
 					struct commit *commit =
 						lookup_commit(result_sha1);
+					if (args.stateless_rpc
+					 && ack == ACK_common
+					 && !(commit->object.flags & COMMON)) {
+						/* We need to replay the have for this object
+						 * on the next RPC request so the peer knows
+						 * it is in common with us.
+						 */
+						const char *hex = sha1_to_hex(result_sha1);
+						packet_buf_write(&req_buf, "have %s\n", hex);
+						state_len = req_buf.len;
+					}
 					mark_common(commit, 0, 1);
 					retval = 0;
 					in_vain = 0;
@@ -339,7 +392,8 @@ static int find_common(int fd[2], unsigned char *result_sha1,
 		}
 	}
 done:
-	packet_write(fd[1], "done\n");
+	packet_buf_write(&req_buf, "done\n");
+	send_request(fd[1], &req_buf);
 	if (args.verbose)
 		fprintf(stderr, "done\n");
 	if (retval != 0) {
@@ -348,6 +402,7 @@ done:
 	}
 	strbuf_release(&req_buf);
 
+	consume_shallow_list(fd[0]);
 	while (flushes || multi_ack) {
 		int ack = get_ack(fd[0], result_sha1);
 		if (ack) {
@@ -672,6 +727,8 @@ static struct ref *do_fetch_pack(int fd[2],
 			 */
 			warning("no common commits");
 
+	if (args.stateless_rpc)
+		packet_flush(fd[1]);
 	if (get_pack(fd, pack_lockfile))
 		die("git fetch-pack: fetch failed.");
 
@@ -742,6 +799,8 @@ int cmd_fetch_pack(int argc, const char **argv, const char *prefix)
 	struct ref *ref = NULL;
 	char *dest = NULL, **heads;
 	int fd[2];
+	char *pack_lockfile = NULL;
+	char **pack_lockfile_ptr = NULL;
 	struct child_process *conn;
 
 	nr_heads = 0;
@@ -791,6 +850,15 @@ int cmd_fetch_pack(int argc, const char **argv, const char *prefix)
 				args.no_progress = 1;
 				continue;
 			}
+			if (!strcmp("--stateless-rpc", arg)) {
+				args.stateless_rpc = 1;
+				continue;
+			}
+			if (!strcmp("--lock-pack", arg)) {
+				args.lock_pack = 1;
+				pack_lockfile_ptr = &pack_lockfile;
+				continue;
+			}
 			usage(fetch_pack_usage);
 		}
 		dest = (char *)arg;
@@ -801,19 +869,27 @@ int cmd_fetch_pack(int argc, const char **argv, const char *prefix)
 	if (!dest)
 		usage(fetch_pack_usage);
 
-	conn = git_connect(fd, (char *)dest, args.uploadpack,
-			   args.verbose ? CONNECT_VERBOSE : 0);
-	if (conn) {
-		get_remote_heads(fd[0], &ref, 0, NULL, 0, NULL);
-
-		ref = fetch_pack(&args, fd, conn, ref, dest, nr_heads, heads, NULL);
-		close(fd[0]);
-		close(fd[1]);
-		if (finish_connect(conn))
-			ref = NULL;
+	if (args.stateless_rpc) {
+		conn = NULL;
+		fd[0] = 0;
+		fd[1] = 1;
 	} else {
-		ref = NULL;
+		conn = git_connect(fd, (char *)dest, args.uploadpack,
+				   args.verbose ? CONNECT_VERBOSE : 0);
+	}
+
+	get_remote_heads(fd[0], &ref, 0, NULL, 0, NULL);
+
+	ref = fetch_pack(&args, fd, conn, ref, dest,
+		nr_heads, heads, pack_lockfile_ptr);
+	if (pack_lockfile) {
+		printf("lock %s\n", pack_lockfile);
+		fflush(stdout);
 	}
+	close(fd[0]);
+	close(fd[1]);
+	if (finish_connect(conn))
+		ref = NULL;
 	ret = !ref;
 
 	if (!ret && nr_heads) {