[PATCH 08/11] Allow for having for_each_ref() list some refs that aren't local

[PATCH 08/11] Allow for having for_each_ref() list some refs that aren't local

[Daniel Barkalow]

This is useful, for example, for listing the refs in a reference
repository during clone, when you don't have your own refs that cover
the objects that are in your alternate repository.

Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
---
 refs.c |   18 ++++++++++++++++++
 refs.h |    9 +++++++++
 2 files changed, 27 insertions(+), 0 deletions(-)

diff --git a/refs.c b/refs.c
index 1b0050e..356396d 100644
--- a/refs.c
+++ b/refs.c
@@ -159,6 +159,8 @@ static struct cached_refs {
 } cached_refs;
 static struct ref_list *current_ref;
 
+static struct ref_list *extra_refs;
+
 static void free_ref_list(struct ref_list *list)
 {
 	struct ref_list *next;
@@ -215,6 +217,17 @@ static void read_packed_refs(FILE *f, struct cached_refs *cached_refs)
 	cached_refs->packed = sort_ref_list(list);
 }
 
+void add_extra_ref(const char *name, const unsigned char *sha1, int flag)
+{
+	extra_refs = add_ref(name, sha1, flag, extra_refs, NULL);
+}
+
+void clear_extra_refs(void)
+{
+	free_ref_list(extra_refs);
+	extra_refs = NULL;
+}
+
 static struct ref_list *get_packed_refs(void)
 {
 	if (!cached_refs.did_packed) {
@@ -535,6 +548,11 @@ static int do_for_each_ref(const char *base, each_ref_fn fn, int trim,
 	struct ref_list *packed = get_packed_refs();
 	struct ref_list *loose = get_loose_refs();
 
+	struct ref_list *extra;
+
+	for (extra = extra_refs; extra; extra = extra->next)
+		retval = do_one_ref(base, fn, trim, cb_data, extra);
+
 	while (packed && loose) {
 		struct ref_list *entry;
 		int cmp = strcmp(packed->name, loose->name);
diff --git a/refs.h b/refs.h
index 06abee1..06ad260 100644
--- a/refs.h
+++ b/refs.h
@@ -24,6 +24,15 @@ extern int for_each_tag_ref(each_ref_fn, void *);
 extern int for_each_branch_ref(each_ref_fn, void *);
 extern int for_each_remote_ref(each_ref_fn, void *);
 
+/*
+ * Extra refs will be listed by for_each_ref() before any actual refs
+ * for the duration of this process or until clear_extra_refs() is
+ * called. Only extra refs added before for_each_ref() is called will
+ * be listed on a given call of for_each_ref().
+ */
+extern void add_extra_ref(const char *refname, const unsigned char *sha1, int flags);
+extern void clear_extra_refs(void);
+
 extern int peel_ref(const char *, unsigned char *);
 
 /** Locks a "refs/" ref returning the lock on success and NULL on failure. **/
-- 
1.5.4.3.327.g614d7.dirty

Re: [PATCH 08/11] Allow for having for_each_ref() list some refs that aren't local

[Johannes Schindelin]

Hi,

On Sat, 8 Mar 2008, Daniel Barkalow wrote:

> This is useful, for example, for listing the refs in a reference 
> repository during clone, when you don't have your own refs that cover 
> the objects that are in your alternate repository.

How about this commit oneline instead:

	for_each_ref(): Allow adding extra refs to be traversed

Hmm?

You can explain in the body that the SHA-1s are not verified locally, so 
that you can add extra refs that do not reference local objects.

Ciao,
Dscho

Re: [PATCH 08/11] Allow for having for_each_ref() list some refs that aren't local

[Daniel Barkalow]

On Sun, 9 Mar 2008, Johannes Schindelin wrote:

> Hi,
> 
> On Sat, 8 Mar 2008, Daniel Barkalow wrote:
> 
> > This is useful, for example, for listing the refs in a reference 
> > repository during clone, when you don't have your own refs that cover 
> > the objects that are in your alternate repository.
> 
> How about this commit oneline instead:
> 
> 	for_each_ref(): Allow adding extra refs to be traversed
> 
> Hmm?
> 
> You can explain in the body that the SHA-1s are not verified locally, so 
> that you can add extra refs that do not reference local objects.

The non-local aspect is actually that the repository's refs doesn't 
actually include them. We actually do have these objects, in an alternate 
repo. Clearly I need a better commit message, with this info not trying to 
fit in the first line.

	-Daniel
*This .sig left intentionally blank*

Re: [PATCH 08/11] Allow for having for_each_ref() list some refs that aren't local

[Junio C Hamano]

Daniel Barkalow <barkalow@iabervon.org> writes:

> This is useful, for example, for listing the refs in a reference
> repository during clone, when you don't have your own refs that cover
> the objects that are in your alternate repository.

I hate to ask a question that I myself do not have a good answer to, but
it is a bit worrysome that using these extra refs would cause for-each-ref
to report different refs with the same name (the alternate's master branch
and this one's master branch are both called refs/heads/master).

I think no caller that reads from for_each_ref() keeps track of names that
it has seen and tries clever things like flagging duplicates as errors,
but it still feels like it is inviting trouble.

Re: [PATCH 08/11] Allow for having for_each_ref() list some refs that aren't local

[Daniel Barkalow]

On Sat, 8 Mar 2008, Junio C Hamano wrote:

> Daniel Barkalow <barkalow@iabervon.org> writes:
> 
> > This is useful, for example, for listing the refs in a reference
> > repository during clone, when you don't have your own refs that cover
> > the objects that are in your alternate repository.
> 
> I hate to ask a question that I myself do not have a good answer to, but
> it is a bit worrysome that using these extra refs would cause for-each-ref
> to report different refs with the same name (the alternate's master branch
> and this one's master branch are both called refs/heads/master).
> 
> I think no caller that reads from for_each_ref() keeps track of names that
> it has seen and tries clever things like flagging duplicates as errors,
> but it still feels like it is inviting trouble.

I considered that, and didn't come to any particular conclusion about 
anything to do about it. In the particular case in this series, there are 
no non-extra refs, and the for_each_ref caller doesn't actually look at 
names at all, so it's really not a problem, but the interface could 
definitely generate odd results. For that matter, non-extra refs are 
sorted by name, and the extra refs are just done first. I think additional 
callers are needed before we can determine in more detail how the 
interface should work, since none of these details matters at all to this 
one.

	-Daniel
*This .sig left intentionally blank*