[PATCH] make shallow repository deepening more network efficient

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

[PATCH] make shallow repository deepening more network efficient

Nicolas Pitre

First of all, I can't find any reason why thin pack generation is
explicitly disabled when dealing with a shallow repository.  The
possible delta base objects are collected from the edge commits which
are always obtained through history walking with the same shallow refs
as the client, Therefore the client is always going to have those base
objects available. So let's remove that restriction.

Then we can make shallow repository deepening much more efficient by
using the remote's unshallowed commits as edge commits to get preferred
base objects for thin pack generation.  On git.git, this makes the data
transfer for the deepening of a shallow repository from depth 1 to depth 2
around 134 KB instead of 3.68 MB.

Signed-off-by: Nicolas Pitre <[hidden email]>

diff --git a/upload-pack.c b/upload-pack.c
index f73e3c9..c77ab71 100644
--- a/upload-pack.c
+++ b/upload-pack.c
@@ -32,6 +32,7 @@ static int no_progress, daemon_mode;
 static int shallow_nr;
 static struct object_array have_obj;
 static struct object_array want_obj;
+static struct object_array extra_edge_obj;
 static unsigned int timeout;
 /* 0 for no sideband,
  * otherwise maximum packet size (up to 65520 bytes).
@@ -135,6 +136,10 @@ static int do_rev_list(int fd, void *create_full_pack)
  if (prepare_revision_walk(&revs))
  die("revision walk setup failed");
  mark_edges_uninteresting(revs.commits, &revs, show_edge);
+ if (use_thin_pack)
+ for (i = 0; i < extra_edge_obj.nr; i++)
+ fprintf(pack_pipe, "-%s\n", sha1_to_hex(
+ extra_edge_obj.objects[i].item->sha1));
  traverse_commit_list(&revs, show_commit, show_object, NULL);
  fflush(pack_pipe);
  fclose(pack_pipe);
@@ -562,7 +567,6 @@ static void receive_needs(void)
  if (!prefixcmp(line, "shallow ")) {
  unsigned char sha1[20];
  struct object *object;
- use_thin_pack = 0;
  if (get_sha1(line + 8, sha1))
  die("invalid shallow line: %s", line);
  object = parse_object(sha1);
@@ -574,7 +578,6 @@ static void receive_needs(void)
  }
  if (!prefixcmp(line, "deepen ")) {
  char *end;
- use_thin_pack = 0;
  depth = strtol(line + 7, &end, 0);
  if (end == line + 7 || depth <= 0)
  die("Invalid deepen: %s", line);
@@ -657,6 +660,7 @@ static void receive_needs(void)
  NULL, &want_obj);
  parents = parents->next;
  }
+ add_object_array(object, NULL, &extra_edge_obj);
  }
  /* make sure commit traversal conforms to client */
  register_shallow(object->sha1);
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to [hidden email]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH] make shallow repository deepening more network efficient

Junio C Hamano
Nicolas Pitre <[hidden email]> writes:

> First of all, I can't find any reason why thin pack generation is
> explicitly disabled when dealing with a shallow repository.  The
> possible delta base objects are collected from the edge commits which
> are always obtained through history walking with the same shallow refs
> as the client, Therefore the client is always going to have those base
> objects available. So let's remove that restriction.
>
> Then we can make shallow repository deepening much more efficient by
> using the remote's unshallowed commits as edge commits to get preferred
> base objects for thin pack generation.  On git.git, this makes the data
> transfer for the deepening of a shallow repository from depth 1 to depth 2
> around 134 KB instead of 3.68 MB.
>
> Signed-off-by: Nicolas Pitre <[hidden email]>

Dscho, this is your code from around ed09aef (support fetching into a
shallow repository, 2006-10-30) and f53514b (allow deepening of a shallow
repository, 2006-10-30).  The latter disables thin pack transfer but the
log does not attempt to justify the change.

Have any comments?

> diff --git a/upload-pack.c b/upload-pack.c
> index f73e3c9..c77ab71 100644
> --- a/upload-pack.c
> +++ b/upload-pack.c
> @@ -32,6 +32,7 @@ static int no_progress, daemon_mode;
>  static int shallow_nr;
>  static struct object_array have_obj;
>  static struct object_array want_obj;
> +static struct object_array extra_edge_obj;
>  static unsigned int timeout;
>  /* 0 for no sideband,
>   * otherwise maximum packet size (up to 65520 bytes).
> @@ -135,6 +136,10 @@ static int do_rev_list(int fd, void *create_full_pack)
>   if (prepare_revision_walk(&revs))
>   die("revision walk setup failed");
>   mark_edges_uninteresting(revs.commits, &revs, show_edge);
> + if (use_thin_pack)
> + for (i = 0; i < extra_edge_obj.nr; i++)
> + fprintf(pack_pipe, "-%s\n", sha1_to_hex(
> + extra_edge_obj.objects[i].item->sha1));
>   traverse_commit_list(&revs, show_commit, show_object, NULL);
>   fflush(pack_pipe);
>   fclose(pack_pipe);
> @@ -562,7 +567,6 @@ static void receive_needs(void)
>   if (!prefixcmp(line, "shallow ")) {
>   unsigned char sha1[20];
>   struct object *object;
> - use_thin_pack = 0;
>   if (get_sha1(line + 8, sha1))
>   die("invalid shallow line: %s", line);
>   object = parse_object(sha1);
> @@ -574,7 +578,6 @@ static void receive_needs(void)
>   }
>   if (!prefixcmp(line, "deepen ")) {
>   char *end;
> - use_thin_pack = 0;
>   depth = strtol(line + 7, &end, 0);
>   if (end == line + 7 || depth <= 0)
>   die("Invalid deepen: %s", line);
> @@ -657,6 +660,7 @@ static void receive_needs(void)
>   NULL, &want_obj);
>   parents = parents->next;
>   }
> + add_object_array(object, NULL, &extra_edge_obj);
>   }
>   /* make sure commit traversal conforms to client */
>   register_shallow(object->sha1);
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to [hidden email]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH] make shallow repository deepening more network efficient

Johannes Schindelin
Hi,

On Sat, 5 Sep 2009, Junio C Hamano wrote:

> Nicolas Pitre <[hidden email]> writes:
>
> > First of all, I can't find any reason why thin pack generation is
> > explicitly disabled when dealing with a shallow repository.  The
> > possible delta base objects are collected from the edge commits which
> > are always obtained through history walking with the same shallow refs
> > as the client, Therefore the client is always going to have those base
> > objects available. So let's remove that restriction.
> >
> > Then we can make shallow repository deepening much more efficient by
> > using the remote's unshallowed commits as edge commits to get preferred
> > base objects for thin pack generation.  On git.git, this makes the data
> > transfer for the deepening of a shallow repository from depth 1 to depth 2
> > around 134 KB instead of 3.68 MB.
> >
> > Signed-off-by: Nicolas Pitre <[hidden email]>
>
> Dscho, this is your code from around ed09aef (support fetching into a
> shallow repository, 2006-10-30) and f53514b (allow deepening of a shallow
> repository, 2006-10-30).  The latter disables thin pack transfer but the
> log does not attempt to justify the change.
>
> Have any comments?

Unfortunately, I do not have any.  I tried to remember, then I tried to
find some documentation, but stopped when I found out that I developed
the code on an iBook which died early Nov 2006.

So no, I do not remember why the change.

Sorry,
Dscho
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to [hidden email]
More majordomo info at  http://vger.kernel.org/majordomo-info.html