Merge branch 'hn/reftable' into seen

The "reftable" backend for the refs API.

* hn/reftable: (28 commits)
  t1404: annotate test cases with REFFILES
  t1401,t2011: parameterize HEAD.lock for REFTABLE
  t1301: document what needs to be done for REFTABLE
  Add "test-tool dump-reftable" command.
  git-prompt: prepare for reftable refs backend
  Reftable support for git-core
  reftable: add dump utility
  reftable: implement stack, a mutable database of reftable files.
  reftable: implement refname validation
  reftable: add merged table view
  reftable: add a heap-based priority queue for reftable records
  reftable: reftable file level tests
  reftable: read reftable files
  reftable: generic interface to tables
  reftable: write reftable files
  reftable: a generic binary tree implementation
  reftable: reading/writing blocks
  Provide zlib's uncompress2 from compat/zlib-compat.c
  reftable: (de)serialization for the polymorphic record type.
  reftable: add blocksource, an abstraction for random access reads
  ...

Documentation/config/extensions.txt

@@ -6,3 +6,12 @@ extensions.objectFormat::
 Note that this setting should only be set by linkgit:git-init[1] or
 linkgit:git-clone[1].  Trying to change it after initialization will not
 work and will produce hard-to-diagnose issues.
++
+extensions.refStorage::
+	Specify the ref storage mechanism to use.  The acceptable values are `files` and
+	`reftable`.  If not specified, `files` is assumed.  It is an error to specify
+	this key unless `core.repositoryFormatVersion` is 1.
++
+Note that this setting should only be set by linkgit:git-init[1] or
+linkgit:git-clone[1].  Trying to change it after initialization will not
+work and will produce hard-to-diagnose issues.

Documentation/technical/repository-version.txt

@@ -100,3 +100,10 @@ If set, by default "git config" reads from both "config" and
 multiple working directory mode, "config" file is shared while
 "config.worktree" is per-working directory (i.e., it's in
 GIT_COMMON_DIR/worktrees/<id>/config.worktree)
+
+==== `refStorage`
+
+Specifies the file format for the ref database. Values are `files`
+(for the traditional packed + loose ref format) and `reftable` for the
+binary reftable format. See https://github.com/google/reftable for
+more information.

Makefile

@@ -256,6 +256,8 @@ all::
 #
 # Define NO_DEFLATE_BOUND if your zlib does not have deflateBound.
 #
+# Define NO_UNCOMPRESS2 if your zlib does not have uncompress2.
+#
 # Define NO_NORETURN if using buggy versions of gcc 4.6+ and profile feedback,
 # as the compiler can crash (http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49299)
 #
@@ -738,6 +740,7 @@ TEST_BUILTINS_OBJS += test-read-cache.o
 TEST_BUILTINS_OBJS += test-read-graph.o
 TEST_BUILTINS_OBJS += test-read-midx.o
 TEST_BUILTINS_OBJS += test-ref-store.o
+TEST_BUILTINS_OBJS += test-reftable.o
 TEST_BUILTINS_OBJS += test-regex.o
 TEST_BUILTINS_OBJS += test-repository.o
 TEST_BUILTINS_OBJS += test-revision-walking.o
@@ -816,6 +819,8 @@ TEST_SHELL_PATH = $(SHELL_PATH)
 
 LIB_FILE = libgit.a
 XDIFF_LIB = xdiff/lib.a
+REFTABLE_LIB = reftable/libreftable.a
+REFTABLE_TEST_LIB = reftable/libreftable_test.a
 
 GENERATED_H += command-list.h
 GENERATED_H += config-list.h
@@ -980,6 +985,7 @@ LIB_OBJS += reflog-walk.o
 LIB_OBJS += refs.o
 LIB_OBJS += refs/debug.o
 LIB_OBJS += refs/files-backend.o
+LIB_OBJS += refs/reftable-backend.o
 LIB_OBJS += refs/iterator.o
 LIB_OBJS += refs/packed-backend.o
 LIB_OBJS += refs/ref-cache.o
@@ -1194,7 +1200,7 @@ THIRD_PARTY_SOURCES += compat/regex/%
 THIRD_PARTY_SOURCES += sha1collisiondetection/%
 THIRD_PARTY_SOURCES += sha1dc/%
 
-GITLIBS = common-main.o $(LIB_FILE) $(XDIFF_LIB)
+GITLIBS = common-main.o $(LIB_FILE) $(XDIFF_LIB) $(REFTABLE_LIB)
 EXTLIBS =
 
 GIT_USER_AGENT = git/$(GIT_VERSION)
@@ -1734,6 +1740,11 @@ ifdef NO_DEFLATE_BOUND
 	BASIC_CFLAGS += -DNO_DEFLATE_BOUND
 endif
 
+ifdef NO_UNCOMPRESS2
+	BASIC_CFLAGS += -DNO_UNCOMPRESS2
+	LIB_OBJS += compat/zlib-uncompress2.o
+endif
+
 ifdef NO_POSIX_GOODIES
 	BASIC_CFLAGS += -DNO_POSIX_GOODIES
 endif
@@ -2438,7 +2449,36 @@ XDIFF_OBJS += xdiff/xutils.o
 .PHONY: xdiff-objs
 xdiff-objs: $(XDIFF_OBJS)
 
+REFTABLE_OBJS += reftable/basics.o
+REFTABLE_OBJS += reftable/error.o
+REFTABLE_OBJS += reftable/block.o
+REFTABLE_OBJS += reftable/blocksource.o
+REFTABLE_OBJS += reftable/iter.o
+REFTABLE_OBJS += reftable/publicbasics.o
+REFTABLE_OBJS += reftable/merged.o
+REFTABLE_OBJS += reftable/pq.o
+REFTABLE_OBJS += reftable/reader.o
+REFTABLE_OBJS += reftable/record.o
+REFTABLE_OBJS += reftable/refname.o
+REFTABLE_OBJS += reftable/generic.o
+REFTABLE_OBJS += reftable/stack.o
+REFTABLE_OBJS += reftable/tree.o
+REFTABLE_OBJS += reftable/writer.o
+
+REFTABLE_TEST_OBJS += reftable/basics_test.o
+REFTABLE_TEST_OBJS += reftable/block_test.o
+REFTABLE_TEST_OBJS += reftable/dump.o
+REFTABLE_TEST_OBJS += reftable/merged_test.o
+REFTABLE_TEST_OBJS += reftable/pq_test.o
+REFTABLE_TEST_OBJS += reftable/record_test.o
+REFTABLE_TEST_OBJS += reftable/readwrite_test.o
+REFTABLE_TEST_OBJS += reftable/refname_test.o
+REFTABLE_TEST_OBJS += reftable/stack_test.o
+REFTABLE_TEST_OBJS += reftable/test_framework.o
+REFTABLE_TEST_OBJS += reftable/tree_test.o
+
 TEST_OBJS := $(patsubst %$X,%.o,$(TEST_PROGRAMS)) $(patsubst %,t/helper/%,$(TEST_BUILTINS_OBJS))
+
 .PHONY: test-objs
 test-objs: $(TEST_OBJS)
 
@@ -2454,6 +2494,8 @@ OBJECTS += $(PROGRAM_OBJS)
 OBJECTS += $(TEST_OBJS)
 OBJECTS += $(XDIFF_OBJS)
 OBJECTS += $(FUZZ_OBJS)
+OBJECTS += $(REFTABLE_OBJS) $(REFTABLE_TEST_OBJS)
+
 ifndef NO_CURL
 	OBJECTS += http.o http-walker.o remote-curl.o
 endif
@@ -2605,6 +2647,12 @@ $(LIB_FILE): $(LIB_OBJS)
 $(XDIFF_LIB): $(XDIFF_OBJS)
 	$(QUIET_AR)$(RM) $@ && $(AR) $(ARFLAGS) $@ $^
 
+$(REFTABLE_LIB): $(REFTABLE_OBJS)
+	$(QUIET_AR)$(RM) $@ && $(AR) $(ARFLAGS) $@ $^
+
+$(REFTABLE_TEST_LIB): $(REFTABLE_TEST_OBJS)
+	$(QUIET_AR)$(RM) $@ && $(AR) $(ARFLAGS) $@ $^
+
 export DEFAULT_EDITOR DEFAULT_PAGER
 
 Documentation/GIT-EXCLUDED-PROGRAMS: FORCE
@@ -2889,7 +2937,7 @@ perf: all
 
 t/helper/test-tool$X: $(patsubst %,t/helper/%,$(TEST_BUILTINS_OBJS))
 
-t/helper/test-%$X: t/helper/test-%.o GIT-LDFLAGS $(GITLIBS)
+t/helper/test-%$X: t/helper/test-%.o GIT-LDFLAGS $(GITLIBS) $(REFTABLE_TEST_LIB)
 	$(QUIET_LINK)$(CC) $(ALL_CFLAGS) -o $@ $(ALL_LDFLAGS) $(filter %.o,$^) $(filter %.a,$^) $(LIBS)
 
 check-sha1:: t/helper/test-tool$X
@@ -3219,7 +3267,7 @@ cocciclean:
 clean: profile-clean coverage-clean cocciclean
 	$(RM) *.res
 	$(RM) $(OBJECTS)
-	$(RM) $(LIB_FILE) $(XDIFF_LIB)
+	$(RM) $(LIB_FILE) $(XDIFF_LIB) $(REFTABLE_LIB) $(REFTABLE_TEST_LIB)
 	$(RM) $(ALL_PROGRAMS) $(SCRIPT_LIB) $(BUILT_INS) git$X
 	$(RM) $(TEST_PROGRAMS)
 	$(RM) $(FUZZ_PROGRAMS)

builtin/clone.c

@@ -1148,7 +1148,7 @@ int cmd_clone(int argc, const char **argv, const char *prefix)
 	}
 
 	init_db(git_dir, real_git_dir, option_template, GIT_HASH_UNKNOWN, NULL,
-		INIT_DB_QUIET);
+		default_ref_storage(), INIT_DB_QUIET);
 
 	if (real_git_dir)
 		git_dir = real_git_dir;
@@ -1299,7 +1299,8 @@ int cmd_clone(int argc, const char **argv, const char *prefix)
 		 * Now that we know what algorithm the remote side is using,
 		 * let's set ours to the same thing.
 		 */
-		initialize_repository_version(hash_algo, 1);
+		initialize_repository_version(hash_algo, 1,
+					      default_ref_storage());
 		repo_set_hash_algo(the_repository, hash_algo);
 
 		mapped_refs = wanted_peer_refs(refs, &remote->fetch);

builtin/init-db.c

@@ -167,12 +167,14 @@ static int needs_work_tree_config(const char *git_dir, const char *work_tree)
 	return 1;
 }
 
-void initialize_repository_version(int hash_algo, int reinit)
+void initialize_repository_version(int hash_algo, int reinit,
+				   const char *ref_storage_format)
 {
 	char repo_version_string[10];
 	int repo_version = GIT_REPO_VERSION;
 
-	if (hash_algo != GIT_HASH_SHA1)
+	if (hash_algo != GIT_HASH_SHA1 ||
+	    !strcmp(ref_storage_format, "reftable"))
 		repo_version = GIT_REPO_VERSION_READ;
 
 	/* This forces creation of new config file */
@@ -225,6 +227,7 @@ static int create_default_files(const char *template_path,
 	is_bare_repository_cfg = init_is_bare_repository || !work_tree;
 	if (init_shared_repository != -1)
 		set_shared_repository(init_shared_repository);
+	the_repository->ref_storage_format = xstrdup(fmt->ref_storage);
 
 	/*
 	 * We would have created the above under user's umask -- under
@@ -234,6 +237,24 @@ static int create_default_files(const char *template_path,
 		adjust_shared_perm(get_git_dir());
 	}
 
+	/*
+	 * Check to see if .git/HEAD exists; this must happen before
+	 * initializing the ref db, because we want to see if there is an
+	 * existing HEAD.
+	 */
+	path = git_path_buf(&buf, "HEAD");
+	reinit = (!access(path, R_OK) ||
+		  readlink(path, junk, sizeof(junk) - 1) != -1);
+
+	/*
+	 * refs/heads is a file when using reftable. We can't reinitialize with
+	 * a reftable because it will overwrite HEAD
+	 */
+	if (reinit && (!strcmp(fmt->ref_storage, "reftable")) ==
+			      is_directory(git_path_buf(&buf, "refs/heads"))) {
+		die("cannot switch ref storage format.");
+	}
+
 	/*
 	 * We need to create a "refs" dir in any case so that older
 	 * versions of git can tell that this is a repository.
@@ -248,9 +269,6 @@ static int create_default_files(const char *template_path,
 	 * Point the HEAD symref to the initial branch with if HEAD does
 	 * not yet exist.
 	 */
-	path = git_path_buf(&buf, "HEAD");
-	reinit = (!access(path, R_OK)
-		  || readlink(path, junk, sizeof(junk)-1) != -1);
 	if (!reinit) {
 		char *ref;
 
@@ -267,7 +285,7 @@ static int create_default_files(const char *template_path,
 		free(ref);
 	}
 
-	initialize_repository_version(fmt->hash_algo, 0);
+	initialize_repository_version(fmt->hash_algo, 0, fmt->ref_storage);
 
 	/* Check filemode trustability */
 	path = git_path_buf(&buf, "config");
@@ -382,7 +400,7 @@ static void validate_hash_algorithm(struct repository_format *repo_fmt, int hash
 
 int init_db(const char *git_dir, const char *real_git_dir,
 	    const char *template_dir, int hash, const char *initial_branch,
-	    unsigned int flags)
+	    const char *ref_storage_format, unsigned int flags)
 {
 	int reinit;
 	int exist_ok = flags & INIT_DB_EXIST_OK;
@@ -421,9 +439,31 @@ int init_db(const char *git_dir, const char *real_git_dir,
 	 * is an attempt to reinitialize new repository with an old tool.
 	 */
 	check_repository_format(&repo_fmt);
+	repo_fmt.ref_storage = xstrdup(ref_storage_format);
 
 	validate_hash_algorithm(&repo_fmt, hash);
 
+	/*
+	 * At this point, the_repository we have in-core does not look
+	 * anything like one that we would see initialized in an already
+	 * working repository after calling setup_git_directory().
+	 *
+	 * Calling repository.c::initialize_the_repository() may have
+	 * prepared the .index .objects and .parsed_objects members, but
+	 * other members like .gitdir, .commondir, etc. have not been
+	 * initialized.
+	 *
+	 * Many API functions assume they are working with the_repository
+	 * that has sensibly been initialized, but because we haven't
+	 * really read from an existing repository, we need to hand-craft
+	 * the necessary members of the structure to get out of this
+	 * chicken-and-egg situation.
+	 *
+	 * For now, we update the hash algorithm member to what the
+	 * validate_hash_algorithm() call decided for us.
+	 */
+	repo_set_hash_algo(the_repository, repo_fmt.hash_algo);
+
 	reinit = create_default_files(template_dir, original_git_dir,
 				      initial_branch, &repo_fmt,
 				      flags & INIT_DB_QUIET);
@@ -454,6 +494,9 @@ int init_db(const char *git_dir, const char *real_git_dir,
 		git_config_set("receive.denyNonFastforwards", "true");
 	}
 
+	if (!strcmp(ref_storage_format, "reftable"))
+		git_config_set("extensions.refStorage", ref_storage_format);
+
 	if (!(flags & INIT_DB_QUIET)) {
 		int len = strlen(git_dir);
 
@@ -527,6 +570,7 @@ static const char *const init_db_usage[] = {
 int cmd_init_db(int argc, const char **argv, const char *prefix)
 {
 	const char *git_dir;
+	const char *ref_storage_format = default_ref_storage();
 	const char *real_git_dir = NULL;
 	const char *work_tree;
 	const char *template_dir = NULL;
@@ -535,15 +579,18 @@ int cmd_init_db(int argc, const char **argv, const char *prefix)
 	const char *initial_branch = NULL;
 	int hash_algo = GIT_HASH_UNKNOWN;
 	const struct option init_db_options[] = {
-		OPT_STRING(0, "template", &template_dir, N_("template-directory"),
-				N_("directory from which templates will be used")),
+		OPT_STRING(0, "template", &template_dir,
+			   N_("template-directory"),
+			   N_("directory from which templates will be used")),
 		OPT_SET_INT(0, "bare", &is_bare_repository_cfg,
-				N_("create a bare repository"), 1),
+			    N_("create a bare repository"), 1),
 		{ OPTION_CALLBACK, 0, "shared", &init_shared_repository,
-			N_("permissions"),
-			N_("specify that the git repository is to be shared amongst several users"),
-			PARSE_OPT_OPTARG | PARSE_OPT_NONEG, shared_callback, 0},
+		  N_("permissions"),
+		  N_("specify that the git repository is to be shared amongst several users"),
+		  PARSE_OPT_OPTARG | PARSE_OPT_NONEG, shared_callback, 0 },
 		OPT_BIT('q', "quiet", &flags, N_("be quiet"), INIT_DB_QUIET),
+		OPT_STRING(0, "ref-storage", &ref_storage_format, N_("backend"),
+			   N_("the ref storage format to use")),
 		OPT_STRING(0, "separate-git-dir", &real_git_dir, N_("gitdir"),
 			   N_("separate git dir from working tree")),
 		OPT_STRING('b', "initial-branch", &initial_branch, N_("name"),
@@ -686,10 +733,11 @@ int cmd_init_db(int argc, const char **argv, const char *prefix)
 	}
 
 	UNLEAK(real_git_dir);
+	UNLEAK(ref_storage_format);
 	UNLEAK(git_dir);
 	UNLEAK(work_tree);
 
 	flags |= INIT_DB_EXIST_OK;
 	return init_db(git_dir, real_git_dir, template_dir, hash_algo,
-		       initial_branch, flags);
+		       initial_branch, ref_storage_format, flags);
 }

builtin/stash.c

@@ -207,10 +207,16 @@ static int get_stash_info(struct stash_info *info, int argc, const char **argv)
 static int do_clear_stash(void)
 {
 	struct object_id obj;
+	int result;
 	if (get_oid(ref_stash, &obj))
 		return 0;
 
-	return delete_ref(NULL, ref_stash, &obj, 0);
+	result = delete_ref(NULL, ref_stash, &obj, 0);
+
+	/* Ignore error; this is necessary for reftable, which keeps reflogs
+	 * even when refs are deleted. */
+	delete_reflog(ref_stash);
+	return result;
 }
 
 static int clear_stash(int argc, const char **argv, const char *prefix)

builtin/worktree.c

@@ -13,6 +13,7 @@
 #include "utf8.h"
 #include "worktree.h"
 #include "quote.h"
+#include "../refs/refs-internal.h"
 
 static const char * const worktree_usage[] = {
 	N_("git worktree add [<options>] <path> [<commit-ish>]"),
@@ -328,9 +329,29 @@ static int add_worktree(const char *path, const char *refname,
 	 * worktree.
 	 */
 	strbuf_reset(&sb);
-	strbuf_addf(&sb, "%s/HEAD", sb_repo.buf);
-	write_file(sb.buf, "%s", oid_to_hex(null_oid()));
-	strbuf_reset(&sb);
+	if (get_main_ref_store(the_repository)->be == &refs_be_reftable) {
+		/* XXX this is cut & paste from reftable_init_db. */
+		strbuf_addf(&sb, "%s/HEAD", sb_repo.buf);
+		write_file(sb.buf, "%s", "ref: refs/heads/.invalid\n");
+		strbuf_reset(&sb);
+
+		strbuf_addf(&sb, "%s/refs", sb_repo.buf);
+		safe_create_dir(sb.buf, 1);
+		strbuf_reset(&sb);
+
+		strbuf_addf(&sb, "%s/refs/heads", sb_repo.buf);
+		write_file(sb.buf, "this repository uses the reftable format");
+		strbuf_reset(&sb);
+
+		strbuf_addf(&sb, "%s/reftable", sb_repo.buf);
+		safe_create_dir(sb.buf, 1);
+		strbuf_reset(&sb);
+	} else {
+		strbuf_addf(&sb, "%s/HEAD", sb_repo.buf);
+		write_file(sb.buf, "%s", oid_to_hex(null_oid()));
+		strbuf_reset(&sb);
+	}
+
 	strbuf_addf(&sb, "%s/commondir", sb_repo.buf);
 	write_file(sb.buf, "../..");