Major submodule rewrite

This replaces the old submodule API with a new extended API that
supports most of the things that can be done with `git submodule`.

include/git2/errors.h

@@ -54,6 +54,7 @@ typedef enum {
 	GITERR_TREE,
 	GITERR_INDEXER,
 	GITERR_SSL,
+	GITERR_SUBMODULE,
 } git_error_t;
 
 /**

include/git2/submodule.h

@@ -20,54 +20,169 @@
  */
 GIT_BEGIN_DECL
 
+/**
+ * Opaque structure representing a submodule.
+ *
+ * Submodule support in libgit2 builds a list of known submodules and keeps
+ * it in the repository.  The list is built from the .gitmodules file, the
+ * .git/config file, the index, and the HEAD tree.  Items in the working
+ * directory that look like submodules (i.e. a git repo) but are not
+ * mentioned in those places won't be tracked.
+ */
+typedef struct git_submodule git_submodule;
+
+/**
+ * Values that could be specified for the update rule of a submodule.
+ *
+ * Use the DEFAULT value if you have altered the update value via
+ * `git_submodule_set_update()` and wish to reset to the original default.
+ */
 typedef enum {
+	GIT_SUBMODULE_UPDATE_DEFAULT = -1,
 	GIT_SUBMODULE_UPDATE_CHECKOUT = 0,
 	GIT_SUBMODULE_UPDATE_REBASE = 1,
-	GIT_SUBMODULE_UPDATE_MERGE = 2
+	GIT_SUBMODULE_UPDATE_MERGE = 2,
+	GIT_SUBMODULE_UPDATE_NONE = 3
 } git_submodule_update_t;
 
+/**
+ * Values that could be specified for how closely to examine the
+ * working directory when getting submodule status.
+ *
+ * Use the DEFUALT value if you have altered the ignore value via
+ * `git_submodule_set_ignore()` and wish to reset to the original value.
+ */
 typedef enum {
-	GIT_SUBMODULE_IGNORE_ALL = 0,       /* never dirty */
-	GIT_SUBMODULE_IGNORE_DIRTY = 1,     /* only dirty if HEAD moved */
-	GIT_SUBMODULE_IGNORE_UNTRACKED = 2, /* dirty if tracked files change */
-	GIT_SUBMODULE_IGNORE_NONE = 3       /* any change or untracked == dirty */
+	GIT_SUBMODULE_IGNORE_DEFAULT = -1,  /* reset to default */
+	GIT_SUBMODULE_IGNORE_NONE = 0,      /* any change or untracked == dirty */
+	GIT_SUBMODULE_IGNORE_UNTRACKED = 1, /* dirty if tracked files change */
+	GIT_SUBMODULE_IGNORE_DIRTY = 2,     /* only dirty if HEAD moved */
+	GIT_SUBMODULE_IGNORE_ALL = 3        /* never dirty */
 } git_submodule_ignore_t;
 
 /**
- * Description of submodule
+ * Status values for submodules.
+ *
+ * One of these values will be returned for the submodule in the index
+ * relative to the HEAD tree, and one will be returned for the submodule in
+ * the working directory relative to the index.  The value can be extracted
+ * from the actual submodule status return value using one of the macros
+ * below (see GIT_SUBMODULE_INDEX_STATUS and GIT_SUBMODULE_WD_STATUS).
+ */
+enum {
+	GIT_SUBMODULE_STATUS_CLEAN = 0,
+	GIT_SUBMODULE_STATUS_ADDED = 1,
+	GIT_SUBMODULE_STATUS_REMOVED = 2,
+	GIT_SUBMODULE_STATUS_REMOVED_TYPE_CHANGE = 3,
+	GIT_SUBMODULE_STATUS_MODIFIED = 4,
+	GIT_SUBMODULE_STATUS_MODIFIED_AHEAD = 5,
+	GIT_SUBMODULE_STATUS_MODIFIED_BEHIND = 6
+};
+
+/**
+ * Return codes for submodule status.
+ *
+ * A combination of these flags (and shifted values of the
+ * GIT_SUBMODULE_STATUS codes above) will be returned to describe the status
+ * of a submodule.
+ *
+ * Submodule info is contained in 4 places: the HEAD tree, the index, config
+ * files (both .git/config and .gitmodules), and the working directory.  Any
+ * or all of those places might be missing information about the submodule
+ * depending on what state the repo is in.
+ *
+ * When you ask for submodule status, we consider all four places and return
+ * a combination of the flags below.  Also, we also compare HEAD to index to
+ * workdir, and return a relative status code (see above) for the
+ * comparisons.  Use the GIT_SUBMODULE_INDEX_STATUS() and
+ * GIT_SUBMODULE_WD_STATUS() macros to extract these status codes from the
+ * results.  As an example, if the submodule exists in the HEAD and does not
+ * exist in the index, then using GIT_SUBMODULE_INDEX_STATUS(st) will return
+ * GIT_SUBMODULE_STATUS_REMOVED.
+ *
+ * The ignore settings for the submodule will control how much status info
+ * you get about the working directory.  For example, with ignore ALL, the
+ * workdir will always show as clean.  With any ignore level below NONE,
+ * you will never get the WD_HAS_UNTRACKED value back.
+ *
+ * The other SUBMODULE_STATUS values you might see are:
+ *
+ * - IN_HEAD means submodule exists in HEAD tree
+ * - IN_INDEX means submodule exists in index
+ * - IN_CONFIG means submodule exists in config
+ * - IN_WD means submodule exists in workdir and looks like a submodule
+ * - WD_CHECKED_OUT means submodule in workdir has .git content
+ * - WD_HAS_UNTRACKED means workdir contains untracked files.  This would
+ *   only ever be returned for ignore value GIT_SUBMODULE_IGNORE_NONE.
+ * - WD_MISSING_COMMITS means workdir repo is out of date and does not
+ *   contain the SHAs from either the index or the HEAD tree
+ */
+#define GIT_SUBMODULE_STATUS_IN_HEAD             (1u << 0)
+#define GIT_SUBMODULE_STATUS_IN_INDEX            (1u << 1)
+#define GIT_SUBMODULE_STATUS_IN_CONFIG           (1u << 2)
+#define GIT_SUBMODULE_STATUS_IN_WD               (1u << 3)
+#define GIT_SUBMODULE_STATUS_INDEX_DATA_OFFSET         (4)
+#define GIT_SUBMODULE_STATUS_WD_DATA_OFFSET            (7)
+#define GIT_SUBMODULE_STATUS_WD_CHECKED_OUT     (1u << 10)
+#define GIT_SUBMODULE_STATUS_WD_HAS_UNTRACKED   (1u << 11)
+#define GIT_SUBMODULE_STATUS_WD_MISSING_COMMITS (1u << 12)
+
+/**
+ * Extract submodule status value for index from status mask.
+ */
+#define GIT_SUBMODULE_INDEX_STATUS(s)  \
+	(((s) >> GIT_SUBMODULE_STATUS_INDEX_DATA_OFFSET) & 0x07)
+
+/**
+ * Extract submodule status value for working directory from status mask.
+ */
+#define GIT_SUBMODULE_WD_STATUS(s)     \
+	(((s) >> GIT_SUBMODULE_STATUS_WD_DATA_OFFSET) & 0x07)
+
+/**
+ * Lookup submodule information by name or path.
+ *
+ * Given either the submodule name or path (they are usually the same), this
+ * returns a structure describing the submodule.
+ *
+ * There are two expected error scenarios:
  *
- * This record describes a submodule found in a repository.  There
- * should be an entry for every submodule found in the HEAD and for
- * every submodule described in .gitmodules.  The fields are as follows:
+ * - The submodule is not mentioned in the HEAD, the index, and the config,
+ *   but does "exist" in the working directory (i.e. there is a subdirectory
+ *   that is a valid self-contained git repo).  In this case, this function
+ *   returns GIT_EEXISTS to indicate the the submodule exists but not in a
+ *   state where a git_submodule can be instantiated.
+ * - The submodule is not mentioned in the HEAD, index, or config and the
+ *   working directory doesn't contain a value git repo at that path.
+ *   There may or may not be anything else at that path, but nothing that
+ *   looks like a submodule.  In this case, this returns GIT_ENOTFOUND.
  *
- * - `name` is the name of the submodule from .gitmodules.
- * - `path` is the path to the submodule from the repo working directory.
- *   It is almost always the same as `name`.
- * - `url` is the url for the submodule.
- * - `oid` is the HEAD SHA1 for the submodule.
- * - `update` is a value from above - see gitmodules(5) update.
- * - `ignore` is a value from above - see gitmodules(5) ignore.
- * - `fetch_recurse` is 0 or 1 - see gitmodules(5) fetchRecurseSubmodules.
- * - `refcount` is for internal use.
+ * The submodule object is owned by the containing repo and will be freed
+ * when the repo is freed.  The caller need not free the submodule.
  *
- * If the submodule has been added to .gitmodules but not yet git added,
- * then the `oid` will be zero.  If the submodule has been deleted, but
- * the delete has not been committed yet, then the `oid` will be set, but
- * the `url` will be NULL.
+ * @param submodule Pointer to submodule description object pointer..
+ * @param repo The repository.
+ * @param name The name of the submodule.  Trailing slashes will be ignored.
+ * @return 0 on success, GIT_ENOTFOUND if submodule does not exist,
+ *         GIT_EEXISTS if submodule exists in working directory only, -1 on
+ *         other errors.
  */
-typedef struct {
-	char *name;
-	char *path;
-	char *url;
-	git_oid oid; /* sha1 of submodule HEAD ref or zero if not committed */
-	git_submodule_update_t update;
-	git_submodule_ignore_t ignore;
-	int fetch_recurse;
-	int refcount;
-} git_submodule;
+GIT_EXTERN(int) git_submodule_lookup(
+	git_submodule **submodule,
+	git_repository *repo,
+	const char *name);
 
 /**
- * Iterate over all submodules of a repository.
+ * Iterate over all tracked submodules of a repository.
+ *
+ * See the note on `git_submodule` above.  This iterates over the tracked
+ * submodules as decribed therein.
+ *
+ * If you are concerned about items in the working directory that look like
+ * submodules but are not tracked, the diff API will generate a diff record
+ * for workdir items that look like submodules but are not tracked, showing
+ * them as added in the workdir.  Also, the status API will treat the entire
+ * subdirectory of a contained git repo as a single GIT_STATUS_WT_NEW item.
  *
  * @param repo The repository
  * @param callback Function to be called with the name of each submodule.
@@ -77,26 +192,286 @@ typedef struct {
  */
 GIT_EXTERN(int) git_submodule_foreach(
 	git_repository *repo,
-	int (*callback)(const char *name, void *payload),
+	int (*callback)(git_submodule *sm, const char *name, void *payload),
 	void *payload);
 
 /**
- * Lookup submodule information by name or path.
+ * Set up a new git submodule for checkout.
  *
- * Given either the submodule name or path (they are usually the same),
- * this returns a structure describing the submodule.  If the submodule
- * does not exist, this will return GIT_ENOTFOUND and set the submodule
- * pointer to NULL.
+ * This does "git submodule add" up to the fetch and checkout of the
+ * submodule contents.  It preps a new submodule, creates an entry in
+ * .gitmodules and creates an empty initialized repository either at the
+ * given path in the working directory or in .git/modules with a gitlink
+ * from the working directory to the new repo.
  *
- * @param submodule Pointer to submodule description object pointer..
- * @param repo The repository.
- * @param name The name of the submodule.  Trailing slashes will be ignored.
- * @return 0 on success, GIT_ENOTFOUND if submodule does not exist, -1 on error
+ * To fully emulate "git submodule add" call this function, then open the
+ * submodule repo and perform the clone step as needed.  Lastly, call
+ * `git_submodule_add_finalize` to wrap up adding the new submodule and
+ * .gitmodules to the index to be ready to commit.
+ *
+ * @param submodule The newly created submodule ready to open for clone
+ * @param repo Superproject repository to contain the new submodule
+ * @param url URL for the submodules remote
+ * @param path Path at which the submodule should be created
+ * @param use_gitlink Should workdir contain a gitlink to the repo in
+ *        .git/modules vs. repo directly in workdir.
+ * @return 0 on success, GIT_EEXISTS if submodule already exists,
+ *         -1 on other errors.
  */
-GIT_EXTERN(int) git_submodule_lookup(
+GIT_EXTERN(int) git_submodule_add_setup(
 	git_submodule **submodule,
 	git_repository *repo,
-	const char *name);
+	const char *url,
+	const char *path,
+	int use_gitlink);
+
+/**
+ * Resolve the setup of a new git submodule.
+ *
+ * This should be called on a submodule once you have called add setup
+ * and done the clone of the submodule.  This adds the .gitmodules file
+ * and the newly cloned submodule to the index to be ready to be committed
+ * (but doesn't actually do the commit).
+ */
+GIT_EXTERN(int) git_submodule_add_finalize(git_submodule *submodule);
+
+/**
+ * Add current submodule HEAD commit to index of superproject.
+ */
+GIT_EXTERN(int) git_submodule_add_to_index(git_submodule *submodule);
+
+/**
+ * Write submodule settings to .gitmodules file.
+ *
+ * This commits any in-memory changes to the submodule to the gitmodules
+ * file on disk.  You may also be interested in `git_submodule_init` which
+ * writes submodule info to ".git/config" (which is better for local changes
+ * to submodule settings) and/or `git_submodule_sync` which writes settings
+ * about remotes to the actual submodule repository.
+ *
+ * @param submodule The submodule to write.
+ * @return 0 on success, <0 on failure.
+ */
+GIT_EXTERN(int) git_submodule_save(git_submodule *submodule);
+
+/**
+ * Get the containing repository for a submodule.
+ *
+ * This returns a pointer to the repository that contains the submodule.
+ * This is a just a reference to the repository that was passed to the
+ * original `git_submodule_lookup` call, so if that repository has been
+ * freed, then this may be a dangling reference.
+ *
+ * @param submodule Pointer to submodule object
+ * @return Pointer to `git_repository`
+ */
+GIT_EXTERN(git_repository *) git_submodule_owner(git_submodule *submodule);
+
+/**
+ * Get the name of submodule.
+ *
+ * @param submodule Pointer to submodule object
+ * @return Pointer to the submodule name
+ */
+GIT_EXTERN(const char *) git_submodule_name(git_submodule *submodule);
+
+/**
+ * Get the path to the submodule.
+ *
+ * The path is almost always the same as the submodule name, but the
+ * two are actually not required to match.
+ *
+ * @param submodule Pointer to submodule object
+ * @return Pointer to the submodule path
+ */
+GIT_EXTERN(const char *) git_submodule_path(git_submodule *submodule);
+
+/**
+ * Get the URL for the submodule.
+ *
+ * @param submodule Pointer to submodule object
+ * @return Pointer to the submodule url
+ */
+GIT_EXTERN(const char *) git_submodule_url(git_submodule *submodule);
+
+/**
+ * Set the URL for the submodule.
+ *
+ * This sets the URL in memory for the submodule. This will be used for
+ * any following submodule actions while this submodule data is in memory.
+ *
+ * After calling this, you may wish to call `git_submodule_save` to write
+ * the changes back to the ".gitmodules" file and `git_submodule_sync` to
+ * write the changes to the checked out submodule repository.
+ *
+ * @param submodule Pointer to the submodule object
+ * @param url URL that should be used for the submodule
+ * @return 0 on success, <0 on failure
+ */
+GIT_EXTERN(int) git_submodule_set_url(git_submodule *submodule, const char *url);
+
+/**
+ * Get the OID for the submodule in the index.
+ *
+ * @param submodule Pointer to submodule object
+ * @return Pointer to git_oid or NULL if submodule is not in index.
+ */
+GIT_EXTERN(const git_oid *) git_submodule_index_oid(git_submodule *submodule);
+
+/**
+ * Get the OID for the submodule in the current HEAD tree.
+ *
+ * @param submodule Pointer to submodule object
+ * @return Pointer to git_oid or NULL if submodule is not in the HEAD.
+ */
+GIT_EXTERN(const git_oid *) git_submodule_head_oid(git_submodule *submodule);
+
+/**
+ * Get the OID for the submodule in the current working directory.
+ *
+ * This returns the OID that corresponds to looking up 'HEAD' in the checked
+ * out submodule.  If there are pending changes in the index or anything
+ * else, this won't notice that.  You should call `git_submodule_status` for
+ * a more complete picture about the state of the working directory.
+ *
+ * @param submodule Pointer to submodule object
+ * @return Pointer to git_oid or NULL if submodule is not checked out.
+ */
+GIT_EXTERN(const git_oid *) git_submodule_wd_oid(git_submodule *submodule);
+
+/**
+ * Get the ignore rule for the submodule.
+ *
+ * There are four ignore values:
+ *
+ *  - **GIT_SUBMODULE_IGNORE_NONE** will consider any change to the contents
+ *    of the submodule from a clean checkout to be dirty, including the
+ *    addition of untracked files.  This is the default if unspecified.
+ *  - **GIT_SUBMODULE_IGNORE_UNTRACKED** examines the contents of the
+ *    working tree (i.e. call `git_status_foreach` on the submodule) but
+ *    UNTRACKED files will not count as making the submodule dirty.
+ *  - **GIT_SUBMODULE_IGNORE_DIRTY** means to only check if the HEAD of the
+ *    submodule has moved for status.  This is fast since it does not need to
+ *    scan the working tree of the submodule at all.
+ *  - **GIT_SUBMODULE_IGNORE_ALL** means not to open the submodule repo.
+ *    The working directory will be consider clean so long as there is a
+ *    checked out version present.
+ */
+GIT_EXTERN(git_submodule_ignore_t) git_submodule_ignore(
+	git_submodule *submodule);
+
+/**
+ * Set the ignore rule for the submodule.
+ *
+ * This sets the ignore rule in memory for the submodule.  This will be used
+ * for any following actions (such as `git_submodule_status`) while the
+ * submodule is in memory.  You should call `git_submodule_save` if you want
+ * to persist the new ignore role.
+ *
+ * Calling this again with GIT_SUBMODULE_IGNORE_DEFAULT or calling
+ * `git_submodule_reload` will revert the rule to the value that was in the
+ * original config.
+ *
+ * @return old value for ignore
+ */
+GIT_EXTERN(git_submodule_ignore_t) git_submodule_set_ignore(
+	git_submodule *submodule,
+	git_submodule_ignore_t ignore);
+
+/**
+ * Get the update rule for the submodule.
+ */
+GIT_EXTERN(git_submodule_update_t) git_submodule_update(
+	git_submodule *submodule);
+
+/**
+ * Set the update rule for the submodule.
+ *
+ * This sets the update rule in memory for the submodule.  You should call
+ * `git_submodule_save` if you want to persist the new update rule.
+ *
+ * Calling this again with GIT_SUBMODULE_UPDATE_DEFAULT or calling
+ * `git_submodule_reload` will revert the rule to the value that was in the
+ * original config.
+ *
+ * @return old value for update
+ */
+GIT_EXTERN(git_submodule_update_t) git_submodule_set_update(
+	git_submodule *submodule,
+	git_submodule_update_t update);
+
+/**
+ * Copy submodule info into ".git/config" file.
+ *
+ * Just like "git submodule init", this copies information about the
+ * submodule into ".git/config".  You can use the accessor functions
+ * above to alter the in-memory git_submodule object and control what
+ * is written to the config, overriding what is in .gitmodules.
+ *
+ * @param submodule The submodule to write into the superproject config
+ * @param overwrite By default, existing entries will not be overwritten,
+ *                  but setting this to true forces them to be updated.
+ * @return 0 on success, <0 on failure.
+ */
+GIT_EXTERN(int) git_submodule_init(git_submodule *submodule, int overwrite);
+
+/**
+ * Copy submodule remote info into submodule repo.
+ *
+ * This copies the information about the submodules URL into the checked out
+ * submodule config, acting like "git submodule sync".  This is useful if
+ * you have altered the URL for the submodule (or it has been altered by a
+ * fetch of upstream changes) and you need to update your local repo.
+ */
+GIT_EXTERN(int) git_submodule_sync(git_submodule *submodule);
+
+/**
+ * Open the repository for a submodule.
+ *
+ * This is a newly opened repository object.  The caller is responsible for
+ * calling `git_repository_free` on it when done.  Multiple calls to this
+ * function will return distinct `git_repository` objects.  This will only
+ * work if the submodule is checked out into the working directory.
+ *
+ * @param subrepo Pointer to the submodule repo which was opened
+ * @param submodule Submodule to be opened
+ * @return 0 on success, <0 if submodule repo could not be opened.
+ */
+GIT_EXTERN(int) git_submodule_open(
+	git_repository **repo,
+	git_submodule *submodule);
+
+/**
+ * Reread submodule info from config, index, and HEAD.
+ *
+ * Call this to reread cached submodule information for this submodule if
+ * you have reason to believe that it has changed.
+ */
+GIT_EXTERN(int) git_submodule_reload(git_submodule *submodule);
+
+/**
+ * Reread all submodule info.
+ *
+ * Call this to reload all cached submodule information for the repo.
+ */
+GIT_EXTERN(int) git_submodule_reload_all(git_repository *repo);
+
+/**
+ * Get the status for a submodule.
+ *
+ * This looks at a submodule and tries to determine the status.  It
+ * will return a combination of the `GIT_SUBMODULE_STATUS` values above.
+ * How deeply it examines the working directory to do this will depend
+ * on the `git_submodule_ignore_t` value for the submodule (which can be
+ * overridden with `git_submodule_set_ignore()`).
+ *
+ * @param status Combination of GIT_SUBMODULE_STATUS values from above.
+ * @param submodule Submodule for which to get status
+ * @return 0 on success, <0 on error
+ */
+GIT_EXTERN(int) git_submodule_status(
+	unsigned int *status,
+	git_submodule *submodule);
 
 /** @} */
 GIT_END_DECL

src/config_file.c

@@ -253,11 +253,17 @@ static int config_set(git_config_file *cfg, const char *name, const char *value)
 		char *tmp = NULL;
 
 		git__free(key);
+
 		if (existing->next != NULL) {
 			giterr_set(GITERR_CONFIG, "Multivar incompatible with simple set");
 			return -1;
 		}
 
+		/* don't update if old and new values already match */
+		if ((!existing->value && !value) ||
+			(existing->value && value && !strcmp(existing->value, value)))
+			return 0;
+
 		if (value) {
 			tmp = git__strdup(value);
 			GITERR_CHECK_ALLOC(tmp);

src/config_file.h

@@ -19,12 +19,24 @@ GIT_INLINE(void) git_config_file_free(git_config_file *cfg)
 	cfg->free(cfg);
 }
 
+GIT_INLINE(int) git_config_file_get_string(
+	const char **out, git_config_file *cfg, const char *name)
+{
+	return cfg->get(cfg, name, out);
+}
+
 GIT_INLINE(int) git_config_file_set_string(
 	git_config_file *cfg, const char *name, const char *value)
 {
 	return cfg->set(cfg, name, value);
 }
 
+GIT_INLINE(int) git_config_file_delete(
+	git_config_file *cfg, const char *name)
+{
+	return cfg->del(cfg, name);
+}
+
 GIT_INLINE(int) git_config_file_foreach(
 	git_config_file *cfg,
 	int (*fn)(const char *key, const char *value, void *data),

src/diff.c

@@ -530,7 +530,7 @@ static int maybe_modified(
 				status = GIT_DELTA_UNMODIFIED;
 			else if (git_submodule_lookup(&sub, diff->repo, nitem->path) < 0)
 				return -1;
-			else if (sub->ignore == GIT_SUBMODULE_IGNORE_ALL)
+			else if (git_submodule_ignore(sub) == GIT_SUBMODULE_IGNORE_ALL)
 				status = GIT_DELTA_UNMODIFIED;
 			else {
 				/* TODO: support other GIT_SUBMODULE_IGNORE values */

src/submodule.c

@@ -17,18 +17,24 @@
 #include "config_file.h"
 #include "config.h"
 #include "repository.h"
+#include "submodule.h"
+#include "tree.h"
+#include "iterator.h"
+
+#define GIT_MODULES_FILE ".gitmodules"
 
 static git_cvar_map _sm_update_map[] = {
 	{GIT_CVAR_STRING, "checkout", GIT_SUBMODULE_UPDATE_CHECKOUT},
 	{GIT_CVAR_STRING, "rebase", GIT_SUBMODULE_UPDATE_REBASE},
-	{GIT_CVAR_STRING, "merge", GIT_SUBMODULE_UPDATE_MERGE}
+	{GIT_CVAR_STRING, "merge", GIT_SUBMODULE_UPDATE_MERGE},
+	{GIT_CVAR_STRING, "none", GIT_SUBMODULE_UPDATE_NONE},
 };
 
 static git_cvar_map _sm_ignore_map[] = {
-	{GIT_CVAR_STRING, "all", GIT_SUBMODULE_IGNORE_ALL},
-	{GIT_CVAR_STRING, "dirty", GIT_SUBMODULE_IGNORE_DIRTY},
+	{GIT_CVAR_STRING, "none", GIT_SUBMODULE_IGNORE_NONE},
 	{GIT_CVAR_STRING, "untracked", GIT_SUBMODULE_IGNORE_UNTRACKED},
-	{GIT_CVAR_STRING, "none", GIT_SUBMODULE_IGNORE_NONE}
+	{GIT_CVAR_STRING, "dirty", GIT_SUBMODULE_IGNORE_DIRTY},
+	{GIT_CVAR_STRING, "all", GIT_SUBMODULE_IGNORE_ALL},
 };
 
 static kh_inline khint_t str_hash_no_trailing_slash(const char *s)
@@ -55,9 +61,725 @@ static kh_inline int str_equal_no_trailing_slash(const char *a, const char *b)
 	return (alen == blen && strncmp(a, b, alen) == 0);
 }
 
-__KHASH_IMPL(str, static kh_inline, const char *, void *, 1, str_hash_no_trailing_slash, str_equal_no_trailing_slash);
+__KHASH_IMPL(
+	str, static kh_inline, const char *, void *, 1,
+	str_hash_no_trailing_slash, str_equal_no_trailing_slash);
+
+static int load_submodule_config(
+	git_repository *repo, bool force);
+static git_config_file *open_gitmodules(
+	git_repository *, bool, const git_oid *);
+static int lookup_head_remote(
+	git_buf *url, git_repository *repo);
+static git_submodule *submodule_lookup_or_create(
+	git_repository *repo, const char *n1, const char *n2);
+static int submodule_update_map(
+	git_repository *repo, git_submodule *sm, const char *key);
+static void submodule_release(
+	git_submodule *sm, int decr);
+static int submodule_load_from_index(
+	git_repository *, const git_index_entry *);
+static int submodule_load_from_head(
+	git_repository *, const char *, const git_oid *);
+static int submodule_load_from_config(
+	const char *, const char *, void *);
+static int submodule_update_config(
+	git_submodule *, const char *, const char *, bool,	bool);
+
+static int submodule_cmp(const void *a, const void *b)
+{
+	return strcmp(((git_submodule *)a)->name, ((git_submodule *)b)->name);
+}
+
+static int submodule_config_key_trunc_puts(git_buf *key, const char *suffix)
+{
+	ssize_t idx = git_buf_rfind(key, '.');
+	git_buf_truncate(key, (size_t)(idx + 1));
+	return git_buf_puts(key, suffix);
+}
+
+/*
+ * PUBLIC APIS
+ */
+
+int git_submodule_lookup(
+	git_submodule **sm_ptr, /* NULL if user only wants to test existence */
+	git_repository *repo,
+	const char *name)       /* trailing slash is allowed */
+{
+	int error;
+	khiter_t pos;
+
+	assert(repo && name);
+
+	if ((error = load_submodule_config(repo, false)) < 0)
+		return error;
+
+	pos = git_strmap_lookup_index(repo->submodules, name);
+
+	if (!git_strmap_valid_index(repo->submodules, pos)) {
+		error = GIT_ENOTFOUND;
+
+		/* check if a plausible submodule exists at path */
+		if (git_repository_workdir(repo)) {
+			git_buf path = GIT_BUF_INIT;
+
+			if (git_buf_joinpath(&path, git_repository_workdir(repo), name) < 0)
+				return -1;
+
+			if (git_path_contains_dir(&path, DOT_GIT))
+				error = GIT_EEXISTS;
+
+			git_buf_free(&path);
+		}
+
+		return error;
+	}
+
+	if (sm_ptr)
+		*sm_ptr = git_strmap_value_at(repo->submodules, pos);
+
+	return 0;
+}
 
-static git_submodule *submodule_alloc(const char *name)
+int git_submodule_foreach(
+	git_repository *repo,
+	int (*callback)(git_submodule *sm, const char *name, void *payload),
+	void *payload)
+{
+	int error;
+	git_submodule *sm;
+	git_vector seen = GIT_VECTOR_INIT;
+	seen._cmp = submodule_cmp;
+
+	assert(repo && callback);
+
+	if ((error = load_submodule_config(repo, false)) < 0)
+		return error;
+
+	git_strmap_foreach_value(repo->submodules, sm, {
+		/* Usually the following will not come into play - it just prevents
+		 * us from issuing a callback twice for a submodule where the name
+		 * and path are not the same.
+		 */
+		if (sm->refcount > 1) {
+			if (git_vector_bsearch(&seen, sm) != GIT_ENOTFOUND)
+				continue;
+			if ((error = git_vector_insert(&seen, sm)) < 0)
+				break;
+		}
+
+		if ((error = callback(sm, sm->name, payload)) < 0)
+			break;
+	});
+
+	git_vector_free(&seen);
+
+	return error;
+}
+
+void git_submodule_config_free(git_repository *repo)
+{
+	git_strmap *smcfg;
+	git_submodule *sm;
+
+	assert(repo);
+
+	smcfg = repo->submodules;
+	repo->submodules = NULL;
+
+	if (smcfg == NULL)
+		return;
+
+	git_strmap_foreach_value(smcfg, sm, {
+		submodule_release(sm,1);
+	});
+	git_strmap_free(smcfg);
+}
+
+int git_submodule_add_setup(
+	git_submodule **submodule,
+	git_repository *repo,
+	const char *url,
+	const char *path,
+	int use_gitlink)
+{
+	int error = 0;
+	git_config_file *mods = NULL;
+	git_submodule *sm;
+	git_buf name = GIT_BUF_INIT, real_url = GIT_BUF_INIT;
+	git_repository_init_options initopt;
+	git_repository *subrepo = NULL;
+
+	assert(repo && url && path);
+
+	/* see if there is already an entry for this submodule */
+
+	if (git_submodule_lookup(&sm, repo, path) < 0)
+		giterr_clear();
+	else {
+		giterr_set(GITERR_SUBMODULE,
+			"Attempt to add a submodule that already exists");
+		return GIT_EEXISTS;
+	}
+
+	/* resolve parameters */
+
+	if (url[0] == '.' && (url[1] == '/' || (url[1] == '.' && url[2] == '/'))) {
+		if (!(error = lookup_head_remote(&real_url, repo)))
+			error = git_path_apply_relative(&real_url, url);
+	} else if (strchr(url, ':') != NULL || url[0] == '/') {
+		error = git_buf_sets(&real_url, url);
+	} else {
+		giterr_set(GITERR_SUBMODULE, "Invalid format for submodule URL");
+		error = -1;
+	}
+	if (error)
+		goto cleanup;
+
+	/* validate and normalize path */
+
+	if (git__prefixcmp(path, git_repository_workdir(repo)) == 0)
+		path += strlen(git_repository_workdir(repo));
+
+	if (git_path_root(path) >= 0) {
+		giterr_set(GITERR_SUBMODULE, "Submodule path must be a relative path");
+		error = -1;
+		goto cleanup;
+	}
+
+	/* update .gitmodules */
+
+	if ((mods = open_gitmodules(repo, true, NULL)) == NULL) {
+		giterr_set(GITERR_SUBMODULE,
+			"Adding submodules to a bare repository is not supported (for now)");
+		return -1;
+	}
+
+	if ((error = git_buf_printf(&name, "submodule.%s.path", path)) < 0 ||
+		(error = git_config_file_set_string(mods, name.ptr, path)) < 0)
+		goto cleanup;
+
+	if ((error = submodule_config_key_trunc_puts(&name, "url")) < 0 ||
+		(error = git_config_file_set_string(mods, name.ptr, real_url.ptr)) < 0)
+		goto cleanup;
+
+	git_buf_clear(&name);
+
+	/* init submodule repository and add origin remote as needed */
+
+	error = git_buf_joinpath(&name, git_repository_workdir(repo), path);
+	if (error < 0)
+		goto cleanup;
+
+	/* New style: sub-repo goes in <repo-dir>/modules/<name>/ with a
+	 * gitlink in the sub-repo workdir directory to that repository
+	 *
+	 * Old style: sub-repo goes directly into repo/<name>/.git/
+	 */
+
+	memset(&initopt, 0, sizeof(initopt));
+	initopt.flags = GIT_REPOSITORY_INIT_MKPATH |
+		GIT_REPOSITORY_INIT_NO_REINIT;
+	initopt.origin_url = real_url.ptr;
+
+	if (git_path_exists(name.ptr) &&
+		git_path_contains(&name, DOT_GIT))
+	{
+		/* repo appears to already exist - reinit? */
+	}
+	else if (use_gitlink) {
+		git_buf repodir = GIT_BUF_INIT;
+
+		error = git_buf_join_n(
+			&repodir, '/', 3, git_repository_path(repo), "modules", path);
+		if (error < 0)
+			goto cleanup;
+
+		initopt.workdir_path = name.ptr;
+		initopt.flags |= GIT_REPOSITORY_INIT_NO_DOTGIT_DIR;
+
+		error = git_repository_init_ext(&subrepo, repodir.ptr, &initopt);
+
+		git_buf_free(&repodir);
+	}
+	else {
+		error = git_repository_init_ext(&subrepo, name.ptr, &initopt);
+	}
+	if (error < 0)
+		goto cleanup;
+
+	/* add submodule to hash and "reload" it */
+
+	if ((sm = submodule_lookup_or_create(repo, path, NULL)) == NULL) {
+		error = -1;
+		goto cleanup;
+	}
+
+	if ((error = submodule_update_map(repo, sm, sm->path)) < 0)
+		goto cleanup;
+
+	if ((error = git_submodule_reload(sm)) < 0)
+		goto cleanup;
+
+	error = git_submodule_init(sm, false);
+
+cleanup:
+	if (submodule != NULL)
+		*submodule = !error ? sm : NULL;
+
+	if (mods != NULL)
+		git_config_file_free(mods);
+	git_repository_free(subrepo);
+	git_buf_free(&real_url);
+	git_buf_free(&name);
+
+	return error;
+}
+
+int git_submodule_add_finalize(git_submodule *sm)
+{
+	int error;
+	git_index *index;
+
+	assert(sm);
+
+	if ((error = git_repository_index__weakptr(&index, sm->owner)) < 0 ||
+		(error = git_index_add(index, GIT_MODULES_FILE, 0)) < 0)
+		return error;
+
+	return git_submodule_add_to_index(sm);
+}
+
+int git_submodule_add_to_index(git_submodule *sm)
+{
+	int error;
+	git_repository *repo, *sm_repo;
+	git_index *index;
+	git_buf path = GIT_BUF_INIT;
+	git_commit *head;
+	git_index_entry entry;
+	struct stat st;
+
+	assert(sm);
+
+	repo = sm->owner;
+
+	if ((error = git_repository_index__weakptr(&index, repo)) < 0 ||
+		(error = git_buf_joinpath(
+			&path, git_repository_workdir(repo), sm->path)) < 0 ||
+		(error = git_submodule_open(&sm_repo, sm)) < 0)
+		goto cleanup;
+
+	/* read stat information for submodule working directory */
+	if (p_stat(path.ptr, &st) < 0) {
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot add submodule without working directory");
+		error = -1;
+		goto cleanup;
+	}
+	git_index__init_entry_from_stat(&st, &entry);
+
+	/* calling git_submodule_open will have set sm->wd_oid if possible */
+	if ((sm->flags & GIT_SUBMODULE_STATUS__WD_OID_VALID) == 0) {
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot add submodule without HEAD to index");
+		error = -1;
+		goto cleanup;
+	}
+	git_oid_cpy(&entry.oid, &sm->wd_oid);
+
+	if ((error = git_commit_lookup(&head, sm_repo, &sm->wd_oid)) < 0)
+		goto cleanup;
+
+	entry.ctime.seconds = git_commit_time(head);
+	entry.ctime.nanoseconds = 0;
+	entry.mtime.seconds = git_commit_time(head);
+	entry.mtime.nanoseconds = 0;
+
+	git_commit_free(head);
+
+	/* now add it */
+	error = git_index_add2(index, &entry);
+
+cleanup:
+	git_repository_free(sm_repo);
+	git_buf_free(&path);
+	return error;
+}
+
+int git_submodule_save(git_submodule *submodule)
+{
+	int error = 0;
+	git_config_file *mods;
+	git_buf key = GIT_BUF_INIT;
+
+	assert(submodule);
+
+	mods = open_gitmodules(submodule->owner, true, NULL);
+	if (!mods) {
+		giterr_set(GITERR_SUBMODULE,
+			"Adding submodules to a bare repository is not supported (for now)");
+		return -1;
+	}
+
+	if ((error = git_buf_printf(&key, "submodule.%s.", submodule->name)) < 0)
+		goto cleanup;
+
+	/* save values for path, url, update, ignore, fetchRecurseSubmodules */
+
+	if ((error = submodule_config_key_trunc_puts(&key, "path")) < 0 ||
+		(error = git_config_file_set_string(mods, key.ptr, submodule->path)) < 0)
+		goto cleanup;
+
+	if ((error = submodule_config_key_trunc_puts(&key, "url")) < 0 ||
+		(error = git_config_file_set_string(mods, key.ptr, submodule->url)) < 0)
+		goto cleanup;
+
+	if (!(error = submodule_config_key_trunc_puts(&key, "update")) &&
+		submodule->update != GIT_SUBMODULE_UPDATE_DEFAULT)
+	{
+		const char *val = (submodule->update == GIT_SUBMODULE_UPDATE_CHECKOUT) ?
+			NULL : _sm_update_map[submodule->update].str_match;
+		error = git_config_file_set_string(mods, key.ptr, val);
+	}
+	if (error < 0)
+		goto cleanup;
+
+	if (!(error = submodule_config_key_trunc_puts(&key, "ignore")) &&
+		submodule->ignore != GIT_SUBMODULE_IGNORE_DEFAULT)
+	{
+		const char *val = (submodule->ignore == GIT_SUBMODULE_IGNORE_NONE) ?
+			NULL : _sm_ignore_map[submodule->ignore].str_match;
+		error = git_config_file_set_string(mods, key.ptr, val);
+	}
+	if (error < 0)
+		goto cleanup;
+
+	if ((error = submodule_config_key_trunc_puts(
+			&key, "fetchRecurseSubmodules")) < 0 ||
+		(error = git_config_file_set_string(
+			mods, key.ptr, submodule->fetch_recurse ? "true" : "false")) < 0)
+		goto cleanup;
+
+	/* update internal defaults */
+
+	submodule->ignore_default = submodule->ignore;
+	submodule->update_default = submodule->update;
+	submodule->flags |= GIT_SUBMODULE_STATUS_IN_CONFIG;
+
+cleanup:
+	if (mods != NULL)
+		git_config_file_free(mods);
+	git_buf_free(&key);
+
+	return error;
+}
+
+git_repository *git_submodule_owner(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->owner;
+}
+
+const char *git_submodule_name(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->name;
+}
+
+const char *git_submodule_path(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->path;
+}
+
+const char *git_submodule_url(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->url;
+}
+
+int git_submodule_set_url(git_submodule *submodule, const char *url)
+{
+	assert(submodule && url);
+
+	git__free(submodule->url);
+
+	submodule->url = git__strdup(url);
+	GITERR_CHECK_ALLOC(submodule->url);
+
+	return 0;
+}
+
+ const git_oid *git_submodule_index_oid(git_submodule *submodule)
+{
+	assert(submodule);
+
+	if (submodule->flags & GIT_SUBMODULE_STATUS__INDEX_OID_VALID)
+		return &submodule->index_oid;
+	else
+		return NULL;
+}
+
+const git_oid *git_submodule_head_oid(git_submodule *submodule)
+{
+	assert(submodule);
+
+	if (submodule->flags & GIT_SUBMODULE_STATUS__HEAD_OID_VALID)
+		return &submodule->head_oid;
+	else
+		return NULL;
+}
+
+const git_oid *git_submodule_wd_oid(git_submodule *submodule)
+{
+	assert(submodule);
+
+	if (!(submodule->flags & GIT_SUBMODULE_STATUS__WD_OID_VALID)) {
+		git_repository *subrepo;
+
+		/* calling submodule open grabs the HEAD OID if possible */
+		if (!git_submodule_open(&subrepo, submodule))
+			git_repository_free(subrepo);
+	}
+
+	if (submodule->flags & GIT_SUBMODULE_STATUS__WD_OID_VALID)
+		return &submodule->wd_oid;
+	else
+		return NULL;
+}
+
+git_submodule_ignore_t git_submodule_ignore(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->ignore;
+}
+
+git_submodule_ignore_t git_submodule_set_ignore(
+	git_submodule *submodule, git_submodule_ignore_t ignore)
+{
+	git_submodule_ignore_t old;
+
+	assert(submodule);
+
+	if (ignore == GIT_SUBMODULE_IGNORE_DEFAULT)
+		ignore = submodule->ignore_default;
+
+	old = submodule->ignore;
+	submodule->ignore = ignore;
+	return old;
+}
+
+git_submodule_update_t git_submodule_update(git_submodule *submodule)
+{
+	assert(submodule);
+	return submodule->update;
+}
+
+git_submodule_update_t git_submodule_set_update(
+	git_submodule *submodule, git_submodule_update_t update)
+{
+	git_submodule_update_t old;
+
+	assert(submodule);
+
+	if (update == GIT_SUBMODULE_UPDATE_DEFAULT)
+		update = submodule->update_default;
+
+	old = submodule->update;
+	submodule->update = update;
+	return old;
+}
+
+int git_submodule_init(git_submodule *submodule, int overwrite)
+{
+	int error;
+
+	/* write "submodule.NAME.url" */
+
+	if (!submodule->url) {
+		giterr_set(GITERR_SUBMODULE,
+			"No URL configured for submodule '%s'", submodule->name);
+		return -1;
+	}
+
+	error = submodule_update_config(
+		submodule, "url", submodule->url, overwrite != 0, false);
+	if (error < 0)
+		return error;
+
+	/* write "submodule.NAME.update" if not default */
+
+	if (submodule->update == GIT_SUBMODULE_UPDATE_CHECKOUT)
+		error = submodule_update_config(
+			submodule, "update", NULL, (overwrite != 0), false);
+	else if (submodule->update != GIT_SUBMODULE_UPDATE_DEFAULT)
+		error = submodule_update_config(
+			submodule, "update",
+			_sm_update_map[submodule->update].str_match,
+			(overwrite != 0), false);
+
+	return error;
+}
+
+int git_submodule_sync(git_submodule *submodule)
+{
+	if (!submodule->url) {
+		giterr_set(GITERR_SUBMODULE,
+			"No URL configured for submodule '%s'", submodule->name);
+		return -1;
+	}
+
+	/* copy URL over to config only if it already exists */
+
+	return submodule_update_config(
+		submodule, "url", submodule->url, true, true);
+}
+
+int git_submodule_open(
+	git_repository **subrepo,
+	git_submodule *submodule)
+{
+	int error;
+	git_buf path = GIT_BUF_INIT;
+	git_repository *repo;
+	const char *workdir;
+
+	assert(submodule && subrepo);
+
+	repo = submodule->owner;
+	workdir = git_repository_workdir(repo);
+
+	if (!workdir) {
+		giterr_set(GITERR_REPOSITORY,
+			"Cannot open submodule repository in a bare repo");
+		return GIT_ENOTFOUND;
+	}
+
+	if ((submodule->flags & GIT_SUBMODULE_STATUS_IN_WD) == 0) {
+		giterr_set(GITERR_REPOSITORY,
+			"Cannot open submodule repository that is not checked out");
+		return GIT_ENOTFOUND;
+	}
+
+	if (git_buf_joinpath(&path, workdir, submodule->path) < 0)
+		return -1;
+
+	error = git_repository_open(subrepo, path.ptr);
+
+	git_buf_free(&path);
+
+	/* if we have opened the submodule successfully, let's grab the HEAD OID */
+	if (!error && !(submodule->flags & GIT_SUBMODULE_STATUS__WD_OID_VALID)) {
+		if (!git_reference_name_to_oid(
+				&submodule->wd_oid, *subrepo, GIT_HEAD_FILE))
+			submodule->flags |= GIT_SUBMODULE_STATUS__WD_OID_VALID;
+		else
+			giterr_clear();
+	}
+
+	return error;
+}
+
+int git_submodule_reload_all(git_repository *repo)
+{
+	assert(repo);
+	return load_submodule_config(repo, true);
+}
+
+int git_submodule_reload(git_submodule *submodule)
+{
+	git_repository *repo;
+	git_index *index;
+	int pos, error;
+	git_tree *head;
+	git_config_file *mods;
+
+	assert(submodule);
+
+	/* refresh index data */
+
+	repo = submodule->owner;
+	if (git_repository_index__weakptr(&index, repo) < 0)
+		return -1;
+
+	pos = git_index_find(index, submodule->path);
+	if (pos >= 0) {
+		git_index_entry *entry = git_index_get(index, pos);
+
+		submodule->flags = submodule->flags &
+			~(GIT_SUBMODULE_STATUS_IN_INDEX |
+			  GIT_SUBMODULE_STATUS__INDEX_OID_VALID);
+
+		if ((error = submodule_load_from_index(repo, entry)) < 0)
+			return error;
+	}
+
+	/* refresh HEAD tree data */
+
+	if (!(error = git_repository_head_tree(&head, repo))) {
+		git_tree_entry *te;
+
+		submodule->flags = submodule->flags &
+			~(GIT_SUBMODULE_STATUS_IN_HEAD |
+			  GIT_SUBMODULE_STATUS__HEAD_OID_VALID);
+
+		if (!(error = git_tree_entry_bypath(&te, head, submodule->path))) {
+			error = submodule_load_from_head(repo, submodule->path, &te->oid);
+
+			git_tree_entry_free(te);
+		}
+		else if (error == GIT_ENOTFOUND) {
+			giterr_clear();
+			error = 0;
+		}
+
+		git_tree_free(head);
+	}
+
+	if (error < 0)
+		return error;
+
+	/* refresh config data */
+
+	if ((mods = open_gitmodules(repo, false, NULL)) != NULL) {
+		git_buf path = GIT_BUF_INIT;
+
+		git_buf_sets(&path, "submodule\\.");
+		git_buf_puts_escape_regex(&path, submodule->name);
+		git_buf_puts(&path, ".*");
+
+		if (git_buf_oom(&path))
+			error = -1;
+		else
+			error = git_config_file_foreach_match(
+				mods, path.ptr, submodule_load_from_config, repo);
+
+		git_buf_free(&path);
+	}
+
+	return error;
+}
+
+int git_submodule_status(
+	unsigned int *status,
+	git_submodule *submodule)
+{
+	assert(status && submodule);
+
+	/* TODO: move status code from below and update */
+
+	*status = 0;
+
+	return 0;
+}
+
+/*
+ * INTERNAL FUNCTIONS
+ */
+
+static git_submodule *submodule_alloc(git_repository *repo, const char *name)
 {
 	git_submodule *sm = git__calloc(1, sizeof(git_submodule));
 	if (sm == NULL)
@@ -69,80 +791,126 @@ static git_submodule *submodule_alloc(const char *name)
 		return NULL;
 	}
 
-	return sm;
+	sm->owner = repo;
+
+	return sm;
+}
+
+static void submodule_release(git_submodule *sm, int decr)
+{
+	if (!sm)
+		return;
+
+	sm->refcount -= decr;
+
+	if (sm->refcount == 0) {
+		if (sm->name != sm->path) {
+			git__free(sm->path);
+			sm->path = NULL;
+		}
+
+		git__free(sm->name);
+		sm->name = NULL;
+
+		git__free(sm->url);
+		sm->url = NULL;
+
+		sm->owner = NULL;
+
+		git__free(sm);
+	}
+}
+
+static git_submodule *submodule_lookup_or_create(
+	git_repository *repo, const char *n1, const char *n2)
+{
+	git_strmap *smcfg = repo->submodules;
+	khiter_t pos;
+	git_submodule *sm;
+
+	assert(n1);
+
+	pos = git_strmap_lookup_index(smcfg, n1);
+
+	if (!git_strmap_valid_index(smcfg, pos) && n2)
+		pos = git_strmap_lookup_index(smcfg, n2);
+
+	if (!git_strmap_valid_index(smcfg, pos))
+		sm = submodule_alloc(repo, n1);
+	else
+		sm = git_strmap_value_at(smcfg, pos);
+
+	return sm;
+}
+
+static int submodule_update_map(
+	git_repository *repo, git_submodule *sm, const char *key)
+{
+	void *old_sm;
+	int error;
+
+	git_strmap_insert2(repo->submodules, key, sm, old_sm, error);
+	if (error < 0) {
+		submodule_release(sm, 0);
+		return -1;
+	}
+
+	sm->refcount++;
+
+	if (old_sm && ((git_submodule *)old_sm) != sm)
+		submodule_release(old_sm, 1);
+
+	return 0;
 }
 
-static void submodule_release(git_submodule *sm, int decr)
+static int submodule_load_from_index(
+	git_repository *repo, const git_index_entry *entry)
 {
-	if (!sm)
-		return;
+	git_submodule *sm = submodule_lookup_or_create(repo, entry->path, NULL);
 
-	sm->refcount -= decr;
+	if (!sm)
+		return -1;
 
-	if (sm->refcount == 0) {
-		if (sm->name != sm->path)
-			git__free(sm->path);
-		git__free(sm->name);
-		git__free(sm->url);
-		git__free(sm);
+	if (sm->flags & GIT_SUBMODULE_STATUS_IN_INDEX) {
+		sm->flags |= GIT_SUBMODULE_STATUS__INDEX_MULTIPLE_ENTRIES;
+		return 0;
 	}
-}
-
-static int submodule_from_entry(
-	git_strmap *smcfg, git_index_entry *entry)
-{
-	git_submodule *sm;
-	void *old_sm;
-	khiter_t pos;
-	int error;
 
-	pos = git_strmap_lookup_index(smcfg, entry->path);
+	sm->flags |= GIT_SUBMODULE_STATUS_IN_INDEX;
 
-	if (git_strmap_valid_index(smcfg, pos))
-		sm = git_strmap_value_at(smcfg, pos);
-	else
-		sm = submodule_alloc(entry->path);
+	git_oid_cpy(&sm->index_oid, &entry->oid);
+	sm->flags |= GIT_SUBMODULE_STATUS__INDEX_OID_VALID;
 
-	git_oid_cpy(&sm->oid, &entry->oid);
+	return submodule_update_map(repo, sm, sm->path);
+}
 
-	if (strcmp(sm->path, entry->path) != 0) {
-		if (sm->path != sm->name) {
-			git__free(sm->path);
-			sm->path = sm->name;
-		}
-		sm->path = git__strdup(entry->path);
-		if (!sm->path)
-			goto fail;
-	}
+static int submodule_load_from_head(
+	git_repository *repo, const char *path, const git_oid *oid)
+{
+	git_submodule *sm = submodule_lookup_or_create(repo, path, NULL);
 
-	git_strmap_insert2(smcfg, sm->path, sm, old_sm, error);
-	if (error < 0)
-		goto fail;
-	sm->refcount++;
+	if (!sm)
+		return -1;
 
-	if (old_sm && ((git_submodule *)old_sm) != sm) {
-		/* TODO: log warning about multiple entrys for same submodule path */
-		submodule_release(old_sm, 1);
-	}
+	sm->flags |= GIT_SUBMODULE_STATUS_IN_HEAD;
 
-	return 0;
+	git_oid_cpy(&sm->head_oid, oid);
+	sm->flags |= GIT_SUBMODULE_STATUS__HEAD_OID_VALID;
 
-fail:
-	submodule_release(sm, 0);
-	return -1;
+	return submodule_update_map(repo, sm, sm->path);
 }
 
-static int submodule_from_config(
+static int submodule_load_from_config(
 	const char *key, const char *value, void *data)
 {
-	git_strmap *smcfg = data;
+	git_repository *repo = data;
+	git_strmap *smcfg = repo->submodules;
 	const char *namestart;
 	const char *property;
 	git_buf name = GIT_BUF_INIT;
 	git_submodule *sm;
 	void *old_sm = NULL;
 	bool is_path;
-	khiter_t pos;
 	int error;
 
 	if (git__prefixcmp(key, "submodule.") != 0)
@@ -153,21 +921,17 @@ static int submodule_from_config(
 	if (property == NULL)
 		return 0;
 	property++;
-	is_path = (strcmp(property, "path") == 0);
+	is_path = (strcasecmp(property, "path") == 0);
 
 	if (git_buf_set(&name, namestart, property - namestart - 1) < 0)
 		return -1;
 
-	pos = git_strmap_lookup_index(smcfg, name.ptr);
-	if (!git_strmap_valid_index(smcfg, pos) && is_path)
-		pos = git_strmap_lookup_index(smcfg, value);
-	if (!git_strmap_valid_index(smcfg, pos))
-		sm = submodule_alloc(name.ptr);
-	else
-		sm = git_strmap_value_at(smcfg, pos);
+	sm = submodule_lookup_or_create(repo, name.ptr, is_path ? value : NULL);
 	if (!sm)
 		goto fail;
 
+	sm->flags |= GIT_SUBMODULE_STATUS_IN_CONFIG;
+
 	if (strcmp(sm->name, name.ptr) != 0) {
 		assert(sm->path == sm->name);
 		sm->name = git_buf_detach(&name);
@@ -177,7 +941,7 @@ static int submodule_from_config(
 			goto fail;
 		sm->refcount++;
 	}
-	else if (is_path && strcmp(sm->path, value) != 0) {
+	else if (is_path && value && strcmp(sm->path, value) != 0) {
 		assert(sm->path == sm->name);
 		sm->path = git__strdup(value);
 		if (sm->path == NULL)
@@ -195,19 +959,23 @@ static int submodule_from_config(
 		submodule_release(old_sm, 1);
 	}
 
+	/* TODO: Look up path in index and if it is present but not a GITLINK
+	 * then this should be deleted (at least to match git's behavior)
+	 */
+
 	if (is_path)
 		return 0;
 
 	/* copy other properties into submodule entry */
-	if (strcmp(property, "url") == 0) {
+	if (strcasecmp(property, "url") == 0) {
 		if (sm->url) {
 			git__free(sm->url);
 			sm->url = NULL;
 		}
-		if ((sm->url = git__strdup(value)) == NULL)
+		if (value != NULL && (sm->url = git__strdup(value)) == NULL)
 			goto fail;
 	}
-	else if (strcmp(property, "update") == 0) {
+	else if (strcasecmp(property, "update") == 0) {
 		int val;
 		if (git_config_lookup_map_value(
 			_sm_update_map, ARRAY_SIZE(_sm_update_map), value, &val) < 0) {
@@ -215,16 +983,16 @@ static int submodule_from_config(
 				"Invalid value for submodule update property: '%s'", value);
 			goto fail;
 		}
-		sm->update = (git_submodule_update_t)val;
+		sm->update_default = sm->update = (git_submodule_update_t)val;
 	}
-	else if (strcmp(property, "fetchRecurseSubmodules") == 0) {
+	else if (strcasecmp(property, "fetchRecurseSubmodules") == 0) {
 		if (git__parse_bool(&sm->fetch_recurse, value) < 0) {
 			giterr_set(GITERR_INVALID,
 				"Invalid value for submodule 'fetchRecurseSubmodules' property: '%s'", value);
 			goto fail;
 		}
 	}
-	else if (strcmp(property, "ignore") == 0) {
+	else if (strcasecmp(property, "ignore") == 0) {
 		int val;
 		if (git_config_lookup_map_value(
 			_sm_ignore_map, ARRAY_SIZE(_sm_ignore_map), value, &val) < 0) {
@@ -232,7 +1000,7 @@ static int submodule_from_config(
 				"Invalid value for submodule ignore property: '%s'", value);
 			goto fail;
 		}
-		sm->ignore = (git_submodule_ignore_t)val;
+		sm->ignore_default = sm->ignore = (git_submodule_ignore_t)val;
 	}
 	/* ignore other unknown submodule properties */
 
@@ -244,144 +1012,471 @@ fail:
 	return -1;
 }
 
-static int load_submodule_config(git_repository *repo)
+static int submodule_load_from_wd_lite(
+	git_submodule *sm, const char *name, void *payload)
+{
+	git_repository *repo = git_submodule_owner(sm);
+	git_buf path = GIT_BUF_INIT;
+
+	GIT_UNUSED(name);
+	GIT_UNUSED(payload);
+
+	if (git_buf_joinpath(&path, git_repository_workdir(repo), sm->path) < 0)
+		return -1;
+
+	if (git_path_isdir(path.ptr))
+		sm->flags |= GIT_SUBMODULE_STATUS__WD_SCANNED;
+
+	if (git_path_contains(&path, DOT_GIT))
+		sm->flags |= GIT_SUBMODULE_STATUS_IN_WD;
+
+	git_buf_free(&path);
+
+	return 0;
+}
+
+static int load_submodule_config_from_index(
+	git_repository *repo, git_oid *gitmodules_oid)
 {
 	int error;
-	git_index *index;
-	unsigned int i, max_i;
-	git_oid gitmodules_oid;
-	git_strmap *smcfg;
-	struct git_config_file *mods = NULL;
+	git_iterator *i;
+	const git_index_entry *entry;
 
-	if (repo->submodules)
-		return 0;
+	if ((error = git_iterator_for_index(&i, repo)) < 0)
+		return error;
 
-	/* submodule data is kept in a hashtable with each submodule stored
-	 * under both its name and its path.  These are usually the same, but
-	 * that is not guaranteed.
-	 */
-	smcfg = git_strmap_alloc();
-	GITERR_CHECK_ALLOC(smcfg);
+	error = git_iterator_current(i, &entry);
 
-	/* scan index for gitmodules (and .gitmodules entry) */
-	if ((error = git_repository_index__weakptr(&index, repo)) < 0)
-		goto cleanup;
-	memset(&gitmodules_oid, 0, sizeof(gitmodules_oid));
-	max_i = git_index_entrycount(index);
+	while (!error && entry != NULL) {
 
-	for (i = 0; i < max_i; i++) {
-		git_index_entry *entry = git_index_get(index, i);
 		if (S_ISGITLINK(entry->mode)) {
-			if ((error = submodule_from_entry(smcfg, entry)) < 0)
-				goto cleanup;
-		}
-		else if (strcmp(entry->path, ".gitmodules") == 0)
-			git_oid_cpy(&gitmodules_oid, &entry->oid);
+			error = submodule_load_from_index(repo, entry);
+			if (error < 0)
+				break;
+		} else if (strcmp(entry->path, GIT_MODULES_FILE) == 0)
+			git_oid_cpy(gitmodules_oid, &entry->oid);
+
+		error = git_iterator_advance(i, &entry);
 	}
 
-	/* load .gitmodules from workdir if it exists */
-	if (git_repository_workdir(repo) != NULL) {
-		/* look in workdir for .gitmodules */
-		git_buf path = GIT_BUF_INIT;
-		if (!git_buf_joinpath(
-				&path, git_repository_workdir(repo), ".gitmodules") &&
-			git_path_isfile(path.ptr))
-		{
-			if (!(error = git_config_file__ondisk(&mods, path.ptr)))
-				error = git_config_file_open(mods);
+	git_iterator_free(i);
+
+	return error;
+}
+
+static int load_submodule_config_from_head(
+	git_repository *repo, git_oid *gitmodules_oid)
+{
+	int error;
+	git_tree *head;
+	git_iterator *i;
+	const git_index_entry *entry;
+
+	if ((error = git_repository_head_tree(&head, repo)) < 0)
+		return error;
+
+	if ((error = git_iterator_for_tree(&i, repo, head)) < 0) {
+		git_tree_free(head);
+		return error;
+	}
+
+	error = git_iterator_current(i, &entry);
+
+	while (!error && entry != NULL) {
+
+		if (S_ISGITLINK(entry->mode)) {
+			error = submodule_load_from_head(repo, entry->path, &entry->oid);
+			if (error < 0)
+				break;
+		} else if (strcmp(entry->path, GIT_MODULES_FILE) == 0 &&
+				   git_oid_iszero(gitmodules_oid))
+			git_oid_cpy(gitmodules_oid, &entry->oid);
+
+		error = git_iterator_advance(i, &entry);
+	}
+
+	git_iterator_free(i);
+	git_tree_free(head);
+
+	return error;
+}
+
+static git_config_file *open_gitmodules(
+	git_repository *repo,
+	bool okay_to_create,
+	const git_oid *gitmodules_oid)
+{
+	const char *workdir = git_repository_workdir(repo);
+	git_buf path = GIT_BUF_INIT;
+	git_config_file *mods = NULL;
+
+	if (workdir != NULL) {
+		if (git_buf_joinpath(&path, workdir, GIT_MODULES_FILE) != 0)
+			return NULL;
+
+		if (okay_to_create || git_path_isfile(path.ptr)) {
+			/* git_config_file__ondisk should only fail if OOM */
+			if (git_config_file__ondisk(&mods, path.ptr) < 0)
+				return NULL;
+
+			/* open should only fail here if the file is malformed */
+			if (git_config_file_open(mods) < 0) {
+				git_config_file_free(mods);
+				mods = NULL;
+			}
 		}
-		git_buf_free(&path);
 	}
 
-	/* load .gitmodules from object cache if not in workdir */
-	if (!error && mods == NULL && !git_oid_iszero(&gitmodules_oid)) {
-		/* TODO: is it worth loading gitmodules from object cache? */
+	if (!mods && gitmodules_oid && !git_oid_iszero(gitmodules_oid)) {
+		/* TODO: Retrieve .gitmodules content from ODB */
+
+		/* Should we actually do this?  Core git does not, but it means you
+		 * can't really get much information about submodules on bare repos.
+		 */
+	}
+
+	return mods;
+}
+
+static int load_submodule_config(git_repository *repo, bool force)
+{
+	int error;
+	git_oid gitmodules_oid;
+	git_buf path = GIT_BUF_INIT;
+	git_config_file *mods = NULL;
+
+	if (repo->submodules && !force)
+		return 0;
+
+	memset(&gitmodules_oid, 0, sizeof(gitmodules_oid));
+
+	/* Submodule data is kept in a hashtable keyed by both name and path.
+	 * These are usually the same, but that is not guaranteed.
+	 */
+	if (!repo->submodules) {
+		repo->submodules = git_strmap_alloc();
+		GITERR_CHECK_ALLOC(repo->submodules);
 	}
 
-	/* process .gitmodules info */
-	if (!error && mods != NULL)
-		error = git_config_file_foreach(mods, submodule_from_config, smcfg);
+	/* add submodule information from index */
+
+	if ((error = load_submodule_config_from_index(repo, &gitmodules_oid)) < 0)
+		goto cleanup;
+
+	/* add submodule information from HEAD */
+
+	if ((error = load_submodule_config_from_head(repo, &gitmodules_oid)) < 0)
+		goto cleanup;
+
+	/* add submodule information from .gitmodules */
+
+	if ((mods = open_gitmodules(repo, false, &gitmodules_oid)) != NULL)
+		error = git_config_file_foreach(mods, submodule_load_from_config, repo);
+
+	if (error != 0)
+		goto cleanup;
+
+	/* shallow scan submodules in work tree */
 
-	/* store submodule config in repo */
-	if (!error)
-		repo->submodules = smcfg;
+	if (!git_repository_is_bare(repo))
+		error = git_submodule_foreach(repo, submodule_load_from_wd_lite, NULL);
 
 cleanup:
+	git_buf_free(&path);
+
 	if (mods != NULL)
 		git_config_file_free(mods);
+
 	if (error)
-		git_strmap_free(smcfg);
+		git_submodule_config_free(repo);
+
 	return error;
 }
 
-void git_submodule_config_free(git_repository *repo)
+static int lookup_head_remote(git_buf *url, git_repository *repo)
 {
-	git_strmap *smcfg = repo->submodules;
-	git_submodule *sm;
+	int error;
+	git_config *cfg;
+	git_reference *head = NULL, *remote = NULL;
+	const char *tgt, *scan;
+	git_buf key = GIT_BUF_INIT;
+
+	/* 1. resolve HEAD -> refs/heads/BRANCH
+	 * 2. lookup config branch.BRANCH.remote -> ORIGIN
+	 * 3. lookup remote.ORIGIN.url
+	 */
 
-	repo->submodules = NULL;
+	if ((error = git_repository_config__weakptr(&cfg, repo)) < 0)
+		return error;
 
-	if (smcfg == NULL)
-		return;
+	if (git_reference_lookup(&head, repo, GIT_HEAD_FILE) < 0) {
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot resolve relative URL when HEAD cannot be resolved");
+		error = GIT_ENOTFOUND;
+		goto cleanup;
+	}
 
-	git_strmap_foreach_value(smcfg, sm, {
-		submodule_release(sm,1);
-	});
-	git_strmap_free(smcfg);
+	if (git_reference_type(head) != GIT_REF_SYMBOLIC) {
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot resolve relative URL when HEAD is not symbolic");
+		error = GIT_ENOTFOUND;
+		goto cleanup;
+	}
+
+	if ((error = git_branch_tracking(&remote, head)) < 0)
+		goto cleanup;
+
+	/* remote should refer to something like refs/remotes/ORIGIN/BRANCH */
+
+	if (git_reference_type(remote) != GIT_REF_SYMBOLIC ||
+		git__prefixcmp(git_reference_target(remote), "refs/remotes/") != 0)
+	{
+		giterr_set(GITERR_SUBMODULE,
+			"Cannot resolve relative URL when HEAD is not symbolic");
+		error = GIT_ENOTFOUND;
+		goto cleanup;
+	}
+
+	scan = tgt = git_reference_target(remote) + strlen("refs/remotes/");
+	while (*scan && (*scan != '/' || (scan > tgt && scan[-1] != '\\')))
+		scan++; /* find non-escaped slash to end ORIGIN name */
+
+	error = git_buf_printf(&key, "remote.%.*s.url", (int)(scan - tgt), tgt);
+	if (error < 0)
+		goto cleanup;
+
+	if ((error = git_config_get_string(&tgt, cfg, key.ptr)) < 0)
+		goto cleanup;
+
+	error = git_buf_sets(url, tgt);
+
+cleanup:
+	git_buf_free(&key);
+	git_reference_free(head);
+	git_reference_free(remote);
+
+	return error;
 }
 
-static int submodule_cmp(const void *a, const void *b)
+static int submodule_update_config(
+	git_submodule *submodule,
+	const char *attr,
+	const char *value,
+	bool overwrite,
+	bool only_existing)
 {
-	return strcmp(((git_submodule *)a)->name, ((git_submodule *)b)->name);
+	int error;
+	git_config *config;
+	git_buf key = GIT_BUF_INIT;
+	const char *old = NULL;
+
+	assert(submodule);
+
+	error = git_repository_config__weakptr(&config, submodule->owner);
+	if (error < 0)
+		return error;
+
+	error = git_buf_printf(&key, "submodule.%s.%s", submodule->name, attr);
+	if (error < 0)
+		goto cleanup;
+
+	if (git_config_get_string(&old, config, key.ptr) < 0)
+		giterr_clear();
+
+	if (!old && only_existing)
+		goto cleanup;
+	if (old && !overwrite)
+		goto cleanup;
+	if ((!old && !value) || (old && value && strcmp(old, value) == 0))
+		goto cleanup;
+
+	if (!value)
+		error = git_config_delete(config, key.ptr);
+	else
+		error = git_config_set_string(config, key.ptr, value);
+
+cleanup:
+	git_buf_free(&key);
+	return error;
 }
 
-int git_submodule_foreach(
-	git_repository *repo,
-	int (*callback)(const char *name, void *payload),
-	void *payload)
+#if 0
+
+static int head_oid_for_submodule(
+	git_oid *oid,
+	git_repository *owner,
+	const char *path)
+{
+	int error = 0;
+	git_oid head_oid;
+	git_tree *head_tree = NULL, *container_tree = NULL;
+	unsigned int pos;
+	const git_tree_entry *entry;
+
+	if (git_reference_name_to_oid(&head_oid, owner, GIT_HEAD_FILE) < 0 ||
+		git_tree_lookup(&head_tree, owner, &head_oid) < 0 ||
+		git_tree_resolve_path(&container_tree, &pos, head_tree, path) < 0 ||
+		(entry = git_tree_entry_byindex(container_tree, pos)) == NULL)
+	{
+		memset(oid, 0, sizeof(*oid));
+		error = GIT_ENOTFOUND;
+	}
+	else {
+		git_oid_cpy(oid, &entry->oid);
+	}
+
+	git_tree_free(head_tree);
+	git_tree_free(container_tree);
+
+	return error;
+}
+
+int git_submodule_status(
+	unsigned int *status,
+	git_oid *head,
+	git_submodule *sm,
+	git_submodule_ignore_t ignore)
 {
 	int error;
-	git_submodule *sm;
-	git_vector seen = GIT_VECTOR_INIT;
-	seen._cmp = submodule_cmp;
+	const char *workdir;
+	git_repository *owner, *sm_repo = NULL;
+	git_oid owner_head, sm_head;
 
-	if ((error = load_submodule_config(repo)) < 0)
-		return error;
+	assert(submodule && status);
 
-	git_strmap_foreach_value(repo->submodules, sm, {
-		/* usually the following will not come into play */
-		if (sm->refcount > 1) {
-			if (git_vector_bsearch(&seen, sm) != GIT_ENOTFOUND)
-				continue;
-			if ((error = git_vector_insert(&seen, sm)) < 0)
-				break;
+	if (head == NULL)
+		head = &sm_head;
+
+	owner   = submodule->owner;
+	workdir = git_repository_workdir(owner);
+
+	if (ignore == GIT_SUBMODULE_IGNORE_DEFAULT)
+		ignore = sm->ignore;
+
+	/* if this is a bare repo or the submodule dir has no .git yet,
+	 * then it is not checked out and we'll just return index data.
+	 */
+	if (!workdir || (sm->flags & GIT_SUBMODULE_FLAG__HAS_DOTGIT) == 0) {
+		*status = GIT_SUBMODULE_STATUS_NOT_CHECKED_OUT;
+
+		if (sm->index_oid_valid)
+			git_oid_cpy(head, &sm->index_oid);
+		else
+			memset(head, 0, sizeof(git_oid));
+
+		if (git_oid_iszero(head)) {
+			if (sm->url)
+				*status = GIT_SUBMODULE_STATUS_NEW_SUBMODULE;
+		} else if (!sm->url) {
+			*status = GIT_SUBMODULE_STATUS_DELETED_SUBMODULE;
 		}
 
-		if ((error = callback(sm->name, payload)) < 0)
-			break;
-	});
+		return 0;
+	}
 
-	git_vector_free(&seen);
+	/* look up submodule path in repo head to find if new or deleted */
+	if ((error = head_oid_for_submodule(&owner_head, owner, sm->path)) < 0) {
+		*status = GIT_SUBMODULE_STATUS_NEW_SUBMODULE;
+		/* ??? */
+	}
+
+	if (ignore == GIT_SUBMODULE_IGNORE_ALL) {
+		*status = GIT_SUBMODULE_STATUS_CLEAN;
+		git_oid_cpy(head, &sm->oid);
+		return 0;
+	}
+
+	if ((error = git_submodule_open(&sm_repo, sm)) < 0)
+		return error;
+
+	if ((error = git_reference_name_to_oid(head, sm_repo, GIT_HEAD_FILE)) < 0)
+		goto cleanup;
+
+	if (ignore == GIT_SUBMODULE_IGNORE_DIRTY &&
+		git_oid_cmp(head, &sm->oid) == 0)
+	{
+		*status = GIT_SUBMODULE_STATUS_CLEAN;
+		return 0;
+	}
+
+	/* look up submodule oid from index in repo to find if new commits or missing commits */
+
+	/* run a short status to find if modified or untracked content */
+
+#define GIT_SUBMODULE_STATUS_NEW_SUBMODULE       (1u << 2)
+#define GIT_SUBMODULE_STATUS_DELETED_SUBMODULE   (1u << 3)
+#define GIT_SUBMODULE_STATUS_NOT_CHECKED_OUT     (1u << 4)
+#define GIT_SUBMODULE_STATUS_NEW_COMMITS         (1u << 5)
+#define GIT_SUBMODULE_STATUS_MISSING_COMMITS     (1u << 6)
+#define GIT_SUBMODULE_STATUS_MODIFIED_CONTENT    (1u << 7)
+#define GIT_SUBMODULE_STATUS_UNTRACKED_CONTENT   (1u << 8)
+
+cleanup:
+	git_repository_free(sm_repo);
+	git_tree_free(owner_tree);
 
 	return error;
 }
 
-int git_submodule_lookup(
-	git_submodule **sm_ptr, /* NULL allowed if user only wants to test */
+int git_submodule_status_for_path(
+	unsigned int *status,
+	git_oid *head,
 	git_repository *repo,
-	const char *name)       /* trailing slash is allowed */
+	const char *submodule_path,
+	git_submodule_ignore_t ignore)
 {
-	khiter_t pos;
+	int error;
+	git_submodule *sm;
+	const char *workdir;
+	git_buf path = GIT_BUF_INIT;
+	git_oid owner_head;
+
+	assert(repo && submodule_path && status);
+
+	if ((error = git_submodule_lookup(&sm, repo, submodule_path)) == 0)
+		return git_submodule_status(status, head, sm, ignore);
+
+	/* if submodule still exists in HEAD, then it is DELETED */
+	if (!(error = head_oid_for_submodule(&owner_head, repo, submodule_path))) {
+		*status = GIT_SUBMODULE_STATUS_DELETED_SUBMODULE;
+		if (head)
+			git_oid_cmp(head, &owner_head);
+		return 0;
+	}
 
-	if (load_submodule_config(repo) < 0)
+	/* submodule was not found - let's see what we can determine about it */
+	workdir = git_repository_workdir(repo);
+
+	if (error != GIT_ENOTFOUND || !workdir) {
+		*status = GIT_SUBMODULE_STATUS_NOT_A_SUBMODULE;
+		return error;
+	}
+
+	giterr_clear();
+	error = 0;
+
+	/* figure out if this is NEW, NOT_CHECKED_OUT, or what */
+	if (git_buf_joinpath(&path, workdir, submodule_path) < 0)
 		return -1;
 
-	pos = git_strmap_lookup_index(repo->submodules, name);
-	if (!git_strmap_valid_index(repo->submodules, pos))
-		return GIT_ENOTFOUND;
+	if (git_path_contains(&path, DOT_GIT)) {
+		git_repository *sm_repo;
 
-	if (sm_ptr)
-		*sm_ptr = git_strmap_value_at(repo->submodules, pos);
+		*status = GIT_SUBMODULE_STATUS_UNTRACKED_SUBMODULE;
 
-	return 0;
+		/* only bother look up head if it was non-NULL */
+		if (head != NULL &&
+			!(error = git_repository_open(&sm_repo, path.ptr)))
+		{
+			error = git_reference_name_to_oid(head, sm_repo, GIT_HEAD_FILE);
+			git_repository_free(sm_repo);
+		}
+	} else
+		*status = GIT_SUBMODULE_STATUS_NOT_A_SUBMODULE;
+
+	git_buf_free(&path);
+
+	return error;
 }
+
+#endif

src/submodule.h

+/*
+ * Copyright (C) 2012 the libgit2 contributors
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+#ifndef INCLUDE_submodule_h__
+#define INCLUDE_submodule_h__
+
+/* Notes:
+ *
+ * Submodule information can be in four places: the index, the config files
+ * (both .git/config and .gitmodules), the HEAD tree, and the working
+ * directory.
+ *
+ * In the index:
+ * - submodule is found by path
+ * - may be missing, present, or of the wrong type
+ * - will have an oid if present
+ *
+ * In the HEAD tree:
+ * - submodule is found by path
+ * - may be missing, present, or of the wrong type
+ * - will have an oid if present
+ *
+ * In the config files:
+ * - submodule is found by submodule "name" which is usually the path
+ * - may be missing or present
+ * - will have a name, path, url, and other properties
+ *
+ * In the working directory:
+ * - submodule is found by path
+ * - may be missing, an empty directory, a checked out directory,
+ *   or of the wrong type
+ * - if checked out, will have a HEAD oid
+ * - if checked out, will have git history that can be used to compare oids
+ * - if checked out, may have modified files and/or untracked files
+ */
+
+/**
+ * Description of submodule
+ *
+ * This record describes a submodule found in a repository.  There should be
+ * an entry for every submodule found in the HEAD and index, and for every
+ * submodule described in .gitmodules.  The fields are as follows:
+ *
+ * - `owner` is the git_repository containing this submodule
+ * - `name` is the name of the submodule from .gitmodules.
+ * - `path` is the path to the submodule from the repo root.  It is almost
+ *    always the same as `name`.
+ * - `url` is the url for the submodule.
+ * - `tree_oid` is the SHA1 for the submodule path in the repo HEAD.
+ * - `index_oid` is the SHA1 for the submodule recorded in the index.
+ * - `workdir_oid` is the SHA1 for the HEAD of the checked out submodule.
+ * - `update` is a git_submodule_update_t value - see gitmodules(5) update.
+ * - `ignore` is a git_submodule_ignore_t value - see gitmodules(5) ignore.
+ * - `fetch_recurse` is 0 or 1 - see gitmodules(5) fetchRecurseSubmodules.
+ * - `refcount` tracks how many hashmap entries there are for this submodule.
+ *   It only comes into play if the name and path of the submodule differ.
+ * - `flags` is for internal use, tracking where this submodule has been
+ *   found (head, index, config, workdir) and other misc info about it.
+ *
+ * If the submodule has been added to .gitmodules but not yet git added,
+ * then the `index_oid` will be valid and zero.  If the submodule has been
+ * deleted, but the delete has not been committed yet, then the `index_oid`
+ * will be set, but the `url` will be NULL.
+ */
+struct git_submodule {
+	git_repository *owner;
+	char *name;
+	char *path; /* important: may point to same string data as "name" */
+	char *url;
+	uint32_t flags;
+	git_oid head_oid;
+	git_oid index_oid;
+	git_oid wd_oid;
+	/* information from config */
+	git_submodule_update_t update;
+	git_submodule_update_t update_default;
+	git_submodule_ignore_t ignore;
+	git_submodule_ignore_t ignore_default;
+	int fetch_recurse;
+	/* internal information */
+	int refcount;
+};
+
+/* Additional flags on top of public GIT_SUBMODULE_STATUS values */
+#define GIT_SUBMODULE_STATUS__WD_SCANNED      (1u << 15)
+#define GIT_SUBMODULE_STATUS__HEAD_OID_VALID  (1u << 16)
+#define GIT_SUBMODULE_STATUS__INDEX_OID_VALID (1u << 17)
+#define GIT_SUBMODULE_STATUS__WD_OID_VALID    (1u << 18)
+#define GIT_SUBMODULE_STATUS__INDEX_MULTIPLE_ENTRIES (1u << 19)
+
+#endif

tests-clar/status/submodules.c

@@ -3,24 +3,17 @@
 #include "path.h"
 #include "posix.h"
 #include "status_helpers.h"
+#include "../submodule/submodule_helpers.h"
 
 static git_repository *g_repo = NULL;
 
 void test_status_submodules__initialize(void)
 {
-	git_buf modpath = GIT_BUF_INIT;
-
 	g_repo = cl_git_sandbox_init("submodules");
 
 	cl_fixture_sandbox("testrepo.git");
 
-	cl_git_pass(git_buf_sets(&modpath, git_repository_workdir(g_repo)));
-	cl_assert(git_path_dirname_r(&modpath, modpath.ptr) >= 0);
-	cl_git_pass(git_buf_joinpath(&modpath, modpath.ptr, "testrepo.git\n"));
-
-	p_rename("submodules/gitmodules", "submodules/.gitmodules");
-	cl_git_append2file("submodules/.gitmodules", modpath.ptr);
-	git_buf_free(&modpath);
+	rewrite_gitmodules(git_repository_workdir(g_repo));
 
 	p_rename("submodules/testrepo/.gitted", "submodules/testrepo/.git");
 }
@@ -28,6 +21,7 @@ void test_status_submodules__initialize(void)
 void test_status_submodules__cleanup(void)
 {
 	cl_git_sandbox_cleanup();
+	cl_fixture_cleanup("testrepo.git");
 }
 
 void test_status_submodules__api(void)
@@ -40,8 +34,8 @@ void test_status_submodules__api(void)
 
 	cl_git_pass(git_submodule_lookup(&sm, g_repo, "testrepo"));
 	cl_assert(sm != NULL);
-	cl_assert_equal_s("testrepo", sm->name);
-	cl_assert_equal_s("testrepo", sm->path);
+	cl_assert_equal_s("testrepo", git_submodule_name(sm));
+	cl_assert_equal_s("testrepo", git_submodule_path(sm));
 }
 
 void test_status_submodules__0(void)

tests-clar/submodule/lookup.c

+#include "clar_libgit2.h"
+#include "submodule_helpers.h"
+#include "posix.h"
+
+static git_repository *g_repo = NULL;
+
+void test_submodule_lookup__initialize(void)
+{
+	g_repo = cl_git_sandbox_init("submod2");
+
+	cl_fixture_sandbox("submod2_target");
+	p_rename("submod2_target/.gitted", "submod2_target/.git");
+
+	/* must create submod2_target before rewrite so prettify will work */
+	rewrite_gitmodules(git_repository_workdir(g_repo));
+	p_rename("submod2/not_submodule/.gitted", "submod2/not_submodule/.git");
+}
+
+void test_submodule_lookup__cleanup(void)
+{
+	cl_git_sandbox_cleanup();
+	cl_fixture_cleanup("submod2_target");
+}
+
+void test_submodule_lookup__simple_lookup(void)
+{
+	git_submodule *sm;
+
+	/* lookup existing */
+	cl_git_pass(git_submodule_lookup(&sm, g_repo, "sm_unchanged"));
+	cl_assert(sm);
+
+	/* lookup pending change in .gitmodules that is not in HEAD */
+	cl_git_pass(git_submodule_lookup(&sm, g_repo, "sm_added_and_uncommited"));
+	cl_assert(sm);
+
+	/* lookup git repo subdir that is not added as submodule */
+	cl_assert(git_submodule_lookup(&sm, g_repo, "not_submodule") == GIT_EEXISTS);
+
+	/* lookup existing directory that is not a submodule */
+	cl_assert(git_submodule_lookup(&sm, g_repo, "just_a_dir") == GIT_ENOTFOUND);
+
+	/* lookup existing file that is not a submodule */
+	cl_assert(git_submodule_lookup(&sm, g_repo, "just_a_file") == GIT_ENOTFOUND);
+
+	/* lookup non-existent item */
+	cl_assert(git_submodule_lookup(&sm, g_repo, "no_such_file") == GIT_ENOTFOUND);
+}
+
+void test_submodule_lookup__accessors(void)
+{
+	git_submodule *sm;
+	const char *oid = "480095882d281ed676fe5b863569520e54a7d5c0";
+
+	cl_git_pass(git_submodule_lookup(&sm, g_repo, "sm_unchanged"));
+	cl_assert(git_submodule_owner(sm) == g_repo);
+	cl_assert_equal_s("sm_unchanged", git_submodule_name(sm));
+	cl_assert(git__suffixcmp(git_submodule_path(sm), "sm_unchanged") == 0);
+	cl_assert(git__suffixcmp(git_submodule_url(sm), "/submod2_target") == 0);
+
+	cl_assert(git_oid_streq(git_submodule_index_oid(sm), oid) == 0);
+	cl_assert(git_oid_streq(git_submodule_head_oid(sm), oid) == 0);
+	cl_assert(git_oid_streq(git_submodule_wd_oid(sm), oid) == 0);
+
+	cl_assert(git_submodule_ignore(sm) == GIT_SUBMODULE_IGNORE_NONE);
+	cl_assert(git_submodule_update(sm) == GIT_SUBMODULE_UPDATE_CHECKOUT);
+
+	cl_git_pass(git_submodule_lookup(&sm, g_repo, "sm_changed_head"));
+	cl_assert_equal_s("sm_changed_head", git_submodule_name(sm));
+
+	cl_assert(git_oid_streq(git_submodule_index_oid(sm), oid) == 0);
+	cl_assert(git_oid_streq(git_submodule_head_oid(sm), oid) == 0);
+	cl_assert(git_oid_streq(git_submodule_wd_oid(sm),
+		"3d9386c507f6b093471a3e324085657a3c2b4247") == 0);
+
+	cl_git_pass(git_submodule_lookup(&sm, g_repo, "sm_added_and_uncommited"));
+	cl_assert_equal_s("sm_added_and_uncommited", git_submodule_name(sm));
+
+	cl_assert(git_oid_streq(git_submodule_index_oid(sm), oid) == 0);
+	cl_assert(git_submodule_head_oid(sm) == NULL);
+	cl_assert(git_oid_streq(git_submodule_wd_oid(sm), oid) == 0);
+
+	cl_git_pass(git_submodule_lookup(&sm, g_repo, "sm_missing_commits"));
+	cl_assert_equal_s("sm_missing_commits", git_submodule_name(sm));
+
+	cl_assert(git_oid_streq(git_submodule_index_oid(sm), oid) == 0);
+	cl_assert(git_oid_streq(git_submodule_head_oid(sm), oid) == 0);
+	cl_assert(git_oid_streq(git_submodule_wd_oid(sm),
+		"5e4963595a9774b90524d35a807169049de8ccad") == 0);
+}
+
+typedef struct {
+	int count;
+} sm_lookup_data;
+
+static int sm_lookup_cb(git_submodule *sm, const char *name, void *payload)
+{
+	sm_lookup_data *data = payload;
+	data->count += 1;
+	cl_assert_equal_s(git_submodule_name(sm), name);
+	return 0;
+}
+
+void test_submodule_lookup__foreach(void)
+{
+	sm_lookup_data data;
+	memset(&data, 0, sizeof(data));
+	cl_git_pass(git_submodule_foreach(g_repo, sm_lookup_cb, &data));
+	cl_assert_equal_i(7, data.count);
+}

tests-clar/submodule/modify.c

+#include "clar_libgit2.h"
+#include "posix.h"
+#include "path.h"
+#include "submodule_helpers.h"
+
+static git_repository *g_repo = NULL;
+
+#define SM_LIBGIT2_URL "https://github.com/libgit2/libgit2.git"
+#define SM_LIBGIT2     "sm_libgit2"
+#define SM_LIBGIT2B    "sm_libgit2b"
+
+void test_submodule_modify__initialize(void)
+{
+	g_repo = cl_git_sandbox_init("submod2");
+
+	cl_fixture_sandbox("submod2_target");
+	p_rename("submod2_target/.gitted", "submod2_target/.git");
+
+	/* must create submod2_target before rewrite so prettify will work */
+	rewrite_gitmodules(git_repository_workdir(g_repo));
+	p_rename("submod2/not_submodule/.gitted", "submod2/not_submodule/.git");
+}
+
+void test_submodule_modify__cleanup(void)
+{
+	cl_git_sandbox_cleanup();
+	cl_fixture_cleanup("submod2_target");
+}
+
+void test_submodule_modify__add(void)
+{
+	git_submodule *sm;
+	git_config *cfg;
+	const char *s;
+
+	/* re-add existing submodule */
+	cl_assert(
+		git_submodule_add_setup(NULL, g_repo, "whatever", "sm_unchanged", 1) ==
+		GIT_EEXISTS );
+
+	/* add a submodule using a gitlink */
+
+	cl_git_pass(
+		git_submodule_add_setup(&sm, g_repo, SM_LIBGIT2_URL, SM_LIBGIT2, 1)
+		);
+
+	cl_assert(git_path_isfile("submod2/" SM_LIBGIT2 "/.git"));
+
+	cl_assert(git_path_isdir("submod2/.git/modules"));
+	cl_assert(git_path_isdir("submod2/.git/modules/" SM_LIBGIT2));
+	cl_assert(git_path_isfile("submod2/.git/modules/" SM_LIBGIT2 "/HEAD"));
+
+	cl_git_pass(git_repository_config(&cfg, g_repo));
+	cl_git_pass(
+		git_config_get_string(&s, cfg, "submodule." SM_LIBGIT2 ".url"));
+	cl_assert_equal_s(s, SM_LIBGIT2_URL);
+	git_config_free(cfg);
+
+	/* add a submodule not using a gitlink */
+
+	cl_git_pass(
+		git_submodule_add_setup(&sm, g_repo, SM_LIBGIT2_URL, SM_LIBGIT2B, 0)
+		);
+
+	cl_assert(git_path_isdir("submod2/" SM_LIBGIT2B "/.git"));
+	cl_assert(git_path_isfile("submod2/" SM_LIBGIT2B "/.git/HEAD"));
+	cl_assert(!git_path_exists("submod2/.git/modules/" SM_LIBGIT2B));
+
+	cl_git_pass(git_repository_config(&cfg, g_repo));
+	cl_git_pass(
+		git_config_get_string(&s, cfg, "submodule." SM_LIBGIT2B ".url"));
+	cl_assert_equal_s(s, SM_LIBGIT2_URL);
+	git_config_free(cfg);
+}
+
+static int delete_one_config(
+	const char *var_name, const char *value, void *payload)
+{
+	git_config *cfg = payload;
+	GIT_UNUSED(value);
+	return git_config_delete(cfg, var_name);
+}
+
+static int init_one_submodule(
+	git_submodule *sm, const char *name, void *payload)
+{
+	GIT_UNUSED(name);
+	GIT_UNUSED(payload);
+	return git_submodule_init(sm, false);
+}
+
+void test_submodule_modify__init(void)
+{
+	git_config *cfg;
+	const char *str;
+
+	/* erase submodule data from .git/config */
+	cl_git_pass(git_repository_config(&cfg, g_repo));
+	cl_git_pass(
+		git_config_foreach_match(cfg, "submodule\\..*", delete_one_config, cfg));
+	git_config_free(cfg);
+
+	/* confirm no submodule data in config */
+	cl_git_pass(git_repository_config(&cfg, g_repo));
+	cl_git_fail(git_config_get_string(&str, cfg, "submodule.sm_unchanged.url"));
+	cl_git_fail(git_config_get_string(&str, cfg, "submodule.sm_changed_head.url"));
+	cl_git_fail(git_config_get_string(&str, cfg, "submodule.sm_added_and_uncommited.url"));
+	git_config_free(cfg);
+
+	/* call init and see that settings are copied */
+	cl_git_pass(git_submodule_foreach(g_repo, init_one_submodule, NULL));
+
+	git_submodule_reload_all(g_repo);
+
+	/* confirm submodule data in config */
+	cl_git_pass(git_repository_config(&cfg, g_repo));
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule.sm_unchanged.url"));
+	cl_assert(git__suffixcmp(str, "/submod2_target") == 0);
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule.sm_changed_head.url"));
+	cl_assert(git__suffixcmp(str, "/submod2_target") == 0);
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule.sm_added_and_uncommited.url"));
+	cl_assert(git__suffixcmp(str, "/submod2_target") == 0);
+	git_config_free(cfg);
+}
+
+static int sync_one_submodule(
+	git_submodule *sm, const char *name, void *payload)
+{
+	GIT_UNUSED(name);
+	GIT_UNUSED(payload);
+	return git_submodule_sync(sm);
+}
+
+void test_submodule_modify__sync(void)
+{
+	git_submodule *sm1, *sm2, *sm3;
+	git_config *cfg;
+	const char *str;
+
+#define SM1 "sm_unchanged"
+#define SM2 "sm_changed_head"
+#define SM3 "sm_added_and_uncommited"
+
+	/* look up some submodules */
+	cl_git_pass(git_submodule_lookup(&sm1, g_repo, SM1));
+	cl_git_pass(git_submodule_lookup(&sm2, g_repo, SM2));
+	cl_git_pass(git_submodule_lookup(&sm3, g_repo, SM3));
+
+	/* At this point, the .git/config URLs for the submodules have
+	 * not be rewritten with the absolute paths (although the
+	 * .gitmodules have.  Let's confirm that they DO NOT match
+	 * yet, then we can do a sync to make them match...
+	 */
+
+	/* check submodule info does not match before sync */
+	cl_git_pass(git_repository_config(&cfg, g_repo));
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule."SM1".url"));
+	cl_assert(strcmp(git_submodule_url(sm1), str) != 0);
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule."SM2".url"));
+	cl_assert(strcmp(git_submodule_url(sm2), str) != 0);
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule."SM3".url"));
+	cl_assert(strcmp(git_submodule_url(sm3), str) != 0);
+	git_config_free(cfg);
+
+	/* sync all the submodules */
+	cl_git_pass(git_submodule_foreach(g_repo, sync_one_submodule, NULL));
+
+	/* check that submodule config is updated */
+	cl_git_pass(git_repository_config(&cfg, g_repo));
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule."SM1".url"));
+	cl_assert_equal_s(git_submodule_url(sm1), str);
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule."SM2".url"));
+	cl_assert_equal_s(git_submodule_url(sm2), str);
+	cl_git_pass(git_config_get_string(&str, cfg, "submodule."SM3".url"));
+	cl_assert_equal_s(git_submodule_url(sm3), str);
+	git_config_free(cfg);
+}
+
+void test_submodule_modify__edit_and_save(void)
+{
+	git_submodule *sm1, *sm2;
+	char *old_url;
+	git_submodule_ignore_t old_ignore;
+	git_submodule_update_t old_update;
+	git_repository *r2;
+
+	cl_git_pass(git_submodule_lookup(&sm1, g_repo, "sm_changed_head"));
+
+	old_url = git__strdup(git_submodule_url(sm1));
+
+	/* modify properties of submodule */
+	cl_git_pass(git_submodule_set_url(sm1, SM_LIBGIT2_URL));
+	old_ignore = git_submodule_set_ignore(sm1, GIT_SUBMODULE_IGNORE_UNTRACKED);
+	old_update = git_submodule_set_update(sm1, GIT_SUBMODULE_UPDATE_REBASE);
+
+	cl_assert_equal_s(SM_LIBGIT2_URL, git_submodule_url(sm1));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_IGNORE_UNTRACKED, (int)git_submodule_ignore(sm1));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_UPDATE_REBASE, (int)git_submodule_update(sm1));
+
+	/* revert without saving (and confirm setters return old value) */
+	cl_git_pass(git_submodule_set_url(sm1, old_url));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_IGNORE_UNTRACKED,
+		(int)git_submodule_set_ignore(sm1, GIT_SUBMODULE_IGNORE_DEFAULT));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_UPDATE_REBASE,
+		(int)git_submodule_set_update(sm1, GIT_SUBMODULE_UPDATE_DEFAULT));
+
+	/* check that revert was successful */
+	cl_assert_equal_s(old_url, git_submodule_url(sm1));
+	cl_assert_equal_i((int)old_ignore, (int)git_submodule_ignore(sm1));
+	cl_assert_equal_i((int)old_update, (int)git_submodule_update(sm1));
+
+	/* modify properties of submodule (again) */
+	cl_git_pass(git_submodule_set_url(sm1, SM_LIBGIT2_URL));
+	git_submodule_set_ignore(sm1, GIT_SUBMODULE_IGNORE_UNTRACKED);
+	git_submodule_set_update(sm1, GIT_SUBMODULE_UPDATE_REBASE);
+
+	/* call save */
+	cl_git_pass(git_submodule_save(sm1));
+
+	/* attempt to "revert" values */
+	git_submodule_set_ignore(sm1, GIT_SUBMODULE_IGNORE_DEFAULT);
+	git_submodule_set_update(sm1, GIT_SUBMODULE_UPDATE_DEFAULT);
+
+	/* but ignore and update should NOT revert because the DEFAULT
+	 * should now be the newly saved value...
+	 */
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_IGNORE_UNTRACKED, (int)git_submodule_ignore(sm1));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_UPDATE_REBASE, (int)git_submodule_update(sm1));
+
+	/* call reload and check that the new values are loaded */
+	cl_git_pass(git_submodule_reload(sm1));
+
+	cl_assert_equal_s(SM_LIBGIT2_URL, git_submodule_url(sm1));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_IGNORE_UNTRACKED, (int)git_submodule_ignore(sm1));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_UPDATE_REBASE, (int)git_submodule_update(sm1));
+
+	/* open a second copy of the repo and compare submodule */
+	cl_git_pass(git_repository_open(&r2, "submod2"));
+	cl_git_pass(git_submodule_lookup(&sm2, r2, "sm_changed_head"));
+
+	cl_assert_equal_s(SM_LIBGIT2_URL, git_submodule_url(sm2));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_IGNORE_UNTRACKED, (int)git_submodule_ignore(sm2));
+	cl_assert_equal_i(
+		(int)GIT_SUBMODULE_UPDATE_REBASE, (int)git_submodule_update(sm2));
+
+	git_repository_free(r2);
+}

tests-clar/submodule/status.c

+#include "clar_libgit2.h"
+#include "posix.h"
+#include "path.h"
+#include "submodule_helpers.h"
+
+static git_repository *g_repo = NULL;
+
+void test_submodule_status__initialize(void)
+{
+	g_repo = cl_git_sandbox_init("submod2");
+
+	cl_fixture_sandbox("submod2_target");
+	p_rename("submod2_target/.gitted", "submod2_target/.git");
+
+	/* must create submod2_target before rewrite so prettify will work */
+	rewrite_gitmodules(git_repository_workdir(g_repo));
+	p_rename("submod2/not_submodule/.gitted", "submod2/not_submodule/.git");
+}
+
+void test_submodule_status__cleanup(void)
+{
+	cl_git_sandbox_cleanup();
+	cl_fixture_cleanup("submod2_target");
+}
+
+void test_submodule_status__unchanged(void)
+{
+	/* make sure it really looks unchanged */
+}
+
+void test_submodule_status__changed(void)
+{
+	/* 4 values of GIT_SUBMODULE_IGNORE to check */
+
+	/* 6 states of change:
+	 * - none, (handled in __unchanged above)
+	 * - dirty workdir file,
+	 * - dirty index,
+	 * - moved head,
+	 * - untracked file,
+	 * - missing commits (i.e. superproject commit is ahead of submodule)
+	 */
+}
+

tests-clar/submodule/submodule_helpers.c

+#include "clar_libgit2.h"
+#include "buffer.h"
+#include "path.h"
+#include "util.h"
+#include "posix.h"
+#include "submodule_helpers.h"
+
+/* rewrite gitmodules -> .gitmodules
+ * rewrite the empty or relative urls inside each module
+ * rename the .gitted directory inside any submodule to .git
+ */
+void rewrite_gitmodules(const char *workdir)
+{
+	git_buf in_f = GIT_BUF_INIT, out_f = GIT_BUF_INIT, path = GIT_BUF_INIT;
+	FILE *in, *out;
+	char line[256];
+
+	cl_git_pass(git_buf_joinpath(&in_f, workdir, "gitmodules"));
+	cl_git_pass(git_buf_joinpath(&out_f, workdir, ".gitmodules"));
+
+	cl_assert((in  = fopen(in_f.ptr, "r")) != NULL);
+	cl_assert((out = fopen(out_f.ptr, "w")) != NULL);
+
+	while (fgets(line, sizeof(line), in) != NULL) {
+		char *scan = line;
+
+		while (*scan == ' ' || *scan == '\t') scan++;
+
+		/* rename .gitted -> .git in submodule directories */
+		if (git__prefixcmp(scan, "path =") == 0) {
+			scan += strlen("path =");
+			while (*scan == ' ') scan++;
+
+			git_buf_joinpath(&path, workdir, scan);
+			git_buf_rtrim(&path);
+			git_buf_joinpath(&path, path.ptr, ".gitted");
+
+			if (!git_buf_oom(&path) && p_access(path.ptr, F_OK) == 0) {
+				git_buf_joinpath(&out_f, workdir, scan);
+				git_buf_rtrim(&out_f);
+				git_buf_joinpath(&out_f, out_f.ptr, ".git");
+
+				if (!git_buf_oom(&out_f))
+					p_rename(path.ptr, out_f.ptr);
+			}
+		}
+
+		/* copy non-"url =" lines verbatim */
+		if (git__prefixcmp(scan, "url =") != 0) {
+			fputs(line, out);
+			continue;
+		}
+
+		/* convert relative URLs in "url =" lines */
+		scan += strlen("url =");
+		while (*scan == ' ') scan++;
+
+		if (*scan == '.') {
+			git_buf_joinpath(&path, workdir, scan);
+			git_buf_rtrim(&path);
+		} else if (!*scan || *scan == '\n') {
+			git_buf_joinpath(&path, workdir, "../testrepo.git");
+		} else {
+			fputs(line, out);
+			continue;
+		}
+
+		git_path_prettify(&path, path.ptr, NULL);
+		git_buf_putc(&path, '\n');
+		cl_assert(!git_buf_oom(&path));
+
+		fwrite(line, scan - line, sizeof(char), out);
+		fputs(path.ptr, out);
+	}
+
+	fclose(in);
+	fclose(out);
+
+	cl_must_pass(p_unlink(in_f.ptr));
+
+	git_buf_free(&in_f);
+	git_buf_free(&out_f);
+	git_buf_free(&path);
+}

tests-clar/submodule/submodule_helpers.h

+extern void rewrite_gitmodules(const char *workdir);
+