2005-07-04 20:26:53 +00:00
|
|
|
#include "cache.h"
|
2005-07-04 22:29:17 +00:00
|
|
|
#include "refs.h"
|
2005-07-04 20:26:53 +00:00
|
|
|
#include "pkt-line.h"
|
2005-10-19 21:27:02 +00:00
|
|
|
#include "commit.h"
|
|
|
|
#include "tag.h"
|
2006-11-01 22:06:23 +00:00
|
|
|
#include "exec_cmd.h"
|
2007-01-23 06:37:33 +00:00
|
|
|
#include "pack.h"
|
2006-11-01 22:06:23 +00:00
|
|
|
#include "sideband.h"
|
2005-07-04 20:26:53 +00:00
|
|
|
|
2005-12-15 06:17:38 +00:00
|
|
|
static int keep_pack;
|
2007-01-25 01:02:15 +00:00
|
|
|
static int transfer_unpack_limit = -1;
|
|
|
|
static int fetch_unpack_limit = -1;
|
2007-01-25 00:47:24 +00:00
|
|
|
static int unpack_limit = 100;
|
2005-07-14 07:08:37 +00:00
|
|
|
static int quiet;
|
2005-08-12 09:08:29 +00:00
|
|
|
static int verbose;
|
2006-03-20 08:21:10 +00:00
|
|
|
static int fetch_all;
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
static int depth;
|
2005-08-12 09:08:29 +00:00
|
|
|
static const char fetch_pack_usage[] =
|
2007-01-23 08:20:17 +00:00
|
|
|
"git-fetch-pack [--all] [--quiet|-q] [--keep|-k] [--thin] [--upload-pack=<git-upload-pack>] [--depth=<n>] [-v] [<host>:]<directory> [<refs>...]";
|
|
|
|
static const char *uploadpack = "git-upload-pack";
|
2005-07-04 20:26:53 +00:00
|
|
|
|
2005-10-19 23:14:34 +00:00
|
|
|
#define COMPLETE (1U << 0)
|
2005-10-28 02:46:27 +00:00
|
|
|
#define COMMON (1U << 1)
|
|
|
|
#define COMMON_REF (1U << 2)
|
|
|
|
#define SEEN (1U << 3)
|
|
|
|
#define POPPED (1U << 4)
|
|
|
|
|
2006-05-25 04:48:34 +00:00
|
|
|
/*
|
|
|
|
* After sending this many "have"s if we do not get any new ACK , we
|
|
|
|
* give up traversing our history.
|
|
|
|
*/
|
|
|
|
#define MAX_IN_VAIN 256
|
|
|
|
|
2006-08-15 17:23:48 +00:00
|
|
|
static struct commit_list *rev_list;
|
|
|
|
static int non_common_revs, multi_ack, use_thin_pack, use_sideband;
|
2005-10-28 02:46:27 +00:00
|
|
|
|
|
|
|
static void rev_list_push(struct commit *commit, int mark)
|
|
|
|
{
|
|
|
|
if (!(commit->object.flags & mark)) {
|
|
|
|
commit->object.flags |= mark;
|
|
|
|
|
|
|
|
if (!(commit->object.parsed))
|
|
|
|
parse_commit(commit);
|
|
|
|
|
|
|
|
insert_by_date(commit, &rev_list);
|
|
|
|
|
|
|
|
if (!(commit->object.flags & COMMON))
|
|
|
|
non_common_revs++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-09-21 05:02:01 +00:00
|
|
|
static int rev_list_insert_ref(const char *path, const unsigned char *sha1, int flag, void *cb_data)
|
2005-10-28 02:46:27 +00:00
|
|
|
{
|
2005-11-02 23:19:13 +00:00
|
|
|
struct object *o = deref_tag(parse_object(sha1), path, 0);
|
2005-10-28 02:46:27 +00:00
|
|
|
|
2006-07-12 03:45:31 +00:00
|
|
|
if (o && o->type == OBJ_COMMIT)
|
2005-10-28 02:46:27 +00:00
|
|
|
rev_list_push((struct commit *)o, SEEN);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
This function marks a rev and its ancestors as common.
|
|
|
|
In some cases, it is desirable to mark only the ancestors (for example
|
|
|
|
when only the server does not yet know that they are common).
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void mark_common(struct commit *commit,
|
|
|
|
int ancestors_only, int dont_parse)
|
|
|
|
{
|
|
|
|
if (commit != NULL && !(commit->object.flags & COMMON)) {
|
|
|
|
struct object *o = (struct object *)commit;
|
|
|
|
|
|
|
|
if (!ancestors_only)
|
|
|
|
o->flags |= COMMON;
|
|
|
|
|
|
|
|
if (!(o->flags & SEEN))
|
|
|
|
rev_list_push(commit, SEEN);
|
|
|
|
else {
|
|
|
|
struct commit_list *parents;
|
|
|
|
|
|
|
|
if (!ancestors_only && !(o->flags & POPPED))
|
|
|
|
non_common_revs--;
|
|
|
|
if (!o->parsed && !dont_parse)
|
|
|
|
parse_commit(commit);
|
|
|
|
|
|
|
|
for (parents = commit->parents;
|
|
|
|
parents;
|
|
|
|
parents = parents->next)
|
|
|
|
mark_common(parents->item, 0, dont_parse);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
Get the next rev to send, ignoring the common.
|
|
|
|
*/
|
|
|
|
|
2006-02-26 15:13:46 +00:00
|
|
|
static const unsigned char* get_rev(void)
|
2005-10-28 02:46:27 +00:00
|
|
|
{
|
|
|
|
struct commit *commit = NULL;
|
|
|
|
|
|
|
|
while (commit == NULL) {
|
|
|
|
unsigned int mark;
|
|
|
|
struct commit_list* parents;
|
|
|
|
|
|
|
|
if (rev_list == NULL || non_common_revs == 0)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
commit = rev_list->item;
|
|
|
|
if (!(commit->object.parsed))
|
|
|
|
parse_commit(commit);
|
|
|
|
commit->object.flags |= POPPED;
|
|
|
|
if (!(commit->object.flags & COMMON))
|
|
|
|
non_common_revs--;
|
|
|
|
|
|
|
|
parents = commit->parents;
|
|
|
|
|
|
|
|
if (commit->object.flags & COMMON) {
|
|
|
|
/* do not send "have", and ignore ancestors */
|
|
|
|
commit = NULL;
|
|
|
|
mark = COMMON | SEEN;
|
|
|
|
} else if (commit->object.flags & COMMON_REF)
|
|
|
|
/* send "have", and ignore ancestors */
|
|
|
|
mark = COMMON | SEEN;
|
|
|
|
else
|
|
|
|
/* send "have", also for its ancestors */
|
|
|
|
mark = SEEN;
|
|
|
|
|
|
|
|
while (parents) {
|
|
|
|
if (!(parents->item->object.flags & SEEN))
|
|
|
|
rev_list_push(parents->item, mark);
|
|
|
|
if (mark & COMMON)
|
|
|
|
mark_common(parents->item, 1, 0);
|
|
|
|
parents = parents->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
rev_list = rev_list->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
return commit->object.sha1;
|
|
|
|
}
|
2005-10-19 23:14:34 +00:00
|
|
|
|
2005-08-12 09:08:29 +00:00
|
|
|
static int find_common(int fd[2], unsigned char *result_sha1,
|
|
|
|
struct ref *refs)
|
2005-07-04 20:26:53 +00:00
|
|
|
{
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
int fetching;
|
2005-10-28 02:46:27 +00:00
|
|
|
int count = 0, flushes = 0, retval;
|
|
|
|
const unsigned char *sha1;
|
2006-05-25 04:48:34 +00:00
|
|
|
unsigned in_vain = 0;
|
|
|
|
int got_continue = 0;
|
2005-10-28 02:46:27 +00:00
|
|
|
|
2006-09-21 04:47:42 +00:00
|
|
|
for_each_ref(rev_list_insert_ref, NULL);
|
2005-07-04 20:26:53 +00:00
|
|
|
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
fetching = 0;
|
|
|
|
for ( ; refs ; refs = refs->next) {
|
2005-08-12 09:08:29 +00:00
|
|
|
unsigned char *remote = refs->old_sha1;
|
2005-10-20 01:28:17 +00:00
|
|
|
struct object *o;
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
|
2005-10-19 23:14:34 +00:00
|
|
|
/*
|
2005-10-20 01:28:17 +00:00
|
|
|
* If that object is complete (i.e. it is an ancestor of a
|
|
|
|
* local ref), we tell them we have it but do not have to
|
|
|
|
* tell them about its ancestors, which they already know
|
|
|
|
* about.
|
2005-10-20 04:55:49 +00:00
|
|
|
*
|
|
|
|
* We use lookup_object here because we are only
|
|
|
|
* interested in the case we *know* the object is
|
|
|
|
* reachable and we have already scanned it.
|
2005-10-20 01:28:17 +00:00
|
|
|
*/
|
2005-10-20 04:55:49 +00:00
|
|
|
if (((o = lookup_object(remote)) != NULL) &&
|
2005-10-28 02:47:07 +00:00
|
|
|
(o->flags & COMPLETE)) {
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
continue;
|
2005-10-19 23:14:34 +00:00
|
|
|
}
|
2005-10-28 02:46:27 +00:00
|
|
|
|
2006-06-21 07:30:21 +00:00
|
|
|
if (!fetching)
|
2006-09-26 15:27:39 +00:00
|
|
|
packet_write(fd[1], "want %s%s%s%s%s%s\n",
|
2006-06-21 07:30:21 +00:00
|
|
|
sha1_to_hex(remote),
|
|
|
|
(multi_ack ? " multi_ack" : ""),
|
2006-09-10 23:27:08 +00:00
|
|
|
(use_sideband == 2 ? " side-band-64k" : ""),
|
|
|
|
(use_sideband == 1 ? " side-band" : ""),
|
2006-09-26 15:27:39 +00:00
|
|
|
(use_thin_pack ? " thin-pack" : ""),
|
|
|
|
" ofs-delta");
|
2006-06-21 07:30:21 +00:00
|
|
|
else
|
|
|
|
packet_write(fd[1], "want %s\n", sha1_to_hex(remote));
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
fetching++;
|
2005-08-12 09:08:29 +00:00
|
|
|
}
|
2006-10-30 19:09:06 +00:00
|
|
|
if (is_repository_shallow())
|
|
|
|
write_shallow_commits(fd[1], 1);
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
if (depth > 0)
|
|
|
|
packet_write(fd[1], "deepen %d", depth);
|
2005-07-04 22:29:17 +00:00
|
|
|
packet_flush(fd[1]);
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
if (!fetching)
|
|
|
|
return 1;
|
2005-10-19 23:14:34 +00:00
|
|
|
|
2006-11-14 06:04:56 +00:00
|
|
|
if (depth > 0) {
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
char line[1024];
|
|
|
|
unsigned char sha1[20];
|
|
|
|
int len;
|
|
|
|
|
|
|
|
while ((len = packet_read_line(fd[0], line, sizeof(line)))) {
|
Mechanical conversion to use prefixcmp()
This mechanically converts strncmp() to use prefixcmp(), but only when
the parameters match specific patterns, so that they can be verified
easily. Leftover from this will be fixed in a separate step, including
idiotic conversions like
if (!strncmp("foo", arg, 3))
=>
if (!(-prefixcmp(arg, "foo")))
This was done by using this script in px.perl
#!/usr/bin/perl -i.bak -p
if (/strncmp\(([^,]+), "([^\\"]*)", (\d+)\)/ && (length($2) == $3)) {
s|strncmp\(([^,]+), "([^\\"]*)", (\d+)\)|prefixcmp($1, "$2")|;
}
if (/strncmp\("([^\\"]*)", ([^,]+), (\d+)\)/ && (length($1) == $3)) {
s|strncmp\("([^\\"]*)", ([^,]+), (\d+)\)|(-prefixcmp($2, "$1"))|;
}
and running:
$ git grep -l strncmp -- '*.c' | xargs perl px.perl
Signed-off-by: Junio C Hamano <junkio@cox.net>
2007-02-20 09:53:29 +00:00
|
|
|
if (!(-prefixcmp(line, "shallow "))) {
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
if (get_sha1_hex(line + 8, sha1))
|
|
|
|
die("invalid shallow line: %s", line);
|
|
|
|
register_shallow(sha1);
|
2006-11-14 06:04:56 +00:00
|
|
|
continue;
|
|
|
|
}
|
Mechanical conversion to use prefixcmp()
This mechanically converts strncmp() to use prefixcmp(), but only when
the parameters match specific patterns, so that they can be verified
easily. Leftover from this will be fixed in a separate step, including
idiotic conversions like
if (!strncmp("foo", arg, 3))
=>
if (!(-prefixcmp(arg, "foo")))
This was done by using this script in px.perl
#!/usr/bin/perl -i.bak -p
if (/strncmp\(([^,]+), "([^\\"]*)", (\d+)\)/ && (length($2) == $3)) {
s|strncmp\(([^,]+), "([^\\"]*)", (\d+)\)|prefixcmp($1, "$2")|;
}
if (/strncmp\("([^\\"]*)", ([^,]+), (\d+)\)/ && (length($1) == $3)) {
s|strncmp\("([^\\"]*)", ([^,]+), (\d+)\)|(-prefixcmp($2, "$1"))|;
}
and running:
$ git grep -l strncmp -- '*.c' | xargs perl px.perl
Signed-off-by: Junio C Hamano <junkio@cox.net>
2007-02-20 09:53:29 +00:00
|
|
|
if (!(-prefixcmp(line, "unshallow "))) {
|
2006-10-30 19:09:53 +00:00
|
|
|
if (get_sha1_hex(line + 10, sha1))
|
|
|
|
die("invalid unshallow line: %s", line);
|
|
|
|
if (!lookup_object(sha1))
|
|
|
|
die("object not found: %s", line);
|
|
|
|
/* make sure that it is parsed as shallow */
|
|
|
|
parse_object(sha1);
|
|
|
|
if (unregister_shallow(sha1))
|
|
|
|
die("no shallow found: %s", line);
|
2006-11-14 06:04:56 +00:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
die("expected shallow/unshallow, got %s", line);
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-10-28 02:46:27 +00:00
|
|
|
flushes = 0;
|
2005-07-04 23:35:13 +00:00
|
|
|
retval = -1;
|
2005-10-28 02:46:27 +00:00
|
|
|
while ((sha1 = get_rev())) {
|
2005-07-04 20:26:53 +00:00
|
|
|
packet_write(fd[1], "have %s\n", sha1_to_hex(sha1));
|
2005-08-12 09:08:29 +00:00
|
|
|
if (verbose)
|
|
|
|
fprintf(stderr, "have %s\n", sha1_to_hex(sha1));
|
2006-05-25 04:48:34 +00:00
|
|
|
in_vain++;
|
2005-07-04 20:26:53 +00:00
|
|
|
if (!(31 & ++count)) {
|
2005-10-28 02:50:26 +00:00
|
|
|
int ack;
|
|
|
|
|
2005-07-04 20:26:53 +00:00
|
|
|
packet_flush(fd[1]);
|
|
|
|
flushes++;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We keep one window "ahead" of the other side, and
|
|
|
|
* will wait for an ACK only on the next one
|
|
|
|
*/
|
|
|
|
if (count == 32)
|
|
|
|
continue;
|
2005-10-28 02:50:26 +00:00
|
|
|
|
|
|
|
do {
|
|
|
|
ack = get_ack(fd[0], result_sha1);
|
|
|
|
if (verbose && ack)
|
|
|
|
fprintf(stderr, "got ack %d %s\n", ack,
|
|
|
|
sha1_to_hex(result_sha1));
|
|
|
|
if (ack == 1) {
|
|
|
|
flushes = 0;
|
|
|
|
multi_ack = 0;
|
|
|
|
retval = 0;
|
|
|
|
goto done;
|
|
|
|
} else if (ack == 2) {
|
|
|
|
struct commit *commit =
|
|
|
|
lookup_commit(result_sha1);
|
|
|
|
mark_common(commit, 0, 1);
|
|
|
|
retval = 0;
|
2006-05-25 04:48:34 +00:00
|
|
|
in_vain = 0;
|
|
|
|
got_continue = 1;
|
2005-10-28 02:50:26 +00:00
|
|
|
}
|
|
|
|
} while (ack);
|
2005-07-04 20:26:53 +00:00
|
|
|
flushes--;
|
2006-05-25 04:48:34 +00:00
|
|
|
if (got_continue && MAX_IN_VAIN < in_vain) {
|
|
|
|
if (verbose)
|
|
|
|
fprintf(stderr, "giving up\n");
|
|
|
|
break; /* give up */
|
|
|
|
}
|
2005-07-04 20:26:53 +00:00
|
|
|
}
|
|
|
|
}
|
2005-10-28 02:50:26 +00:00
|
|
|
done:
|
2005-07-04 23:35:13 +00:00
|
|
|
packet_write(fd[1], "done\n");
|
2005-08-12 09:08:29 +00:00
|
|
|
if (verbose)
|
|
|
|
fprintf(stderr, "done\n");
|
2005-10-28 02:50:26 +00:00
|
|
|
if (retval != 0) {
|
|
|
|
multi_ack = 0;
|
2005-10-28 02:46:27 +00:00
|
|
|
flushes++;
|
2005-10-28 02:50:26 +00:00
|
|
|
}
|
|
|
|
while (flushes || multi_ack) {
|
|
|
|
int ack = get_ack(fd[0], result_sha1);
|
|
|
|
if (ack) {
|
2005-08-12 09:08:29 +00:00
|
|
|
if (verbose)
|
2005-10-28 02:50:26 +00:00
|
|
|
fprintf(stderr, "got ack (%d) %s\n", ack,
|
|
|
|
sha1_to_hex(result_sha1));
|
|
|
|
if (ack == 1)
|
|
|
|
return 0;
|
|
|
|
multi_ack = 1;
|
|
|
|
continue;
|
2005-08-12 09:08:29 +00:00
|
|
|
}
|
2005-10-28 02:50:26 +00:00
|
|
|
flushes--;
|
2005-07-04 20:26:53 +00:00
|
|
|
}
|
2005-07-04 23:35:13 +00:00
|
|
|
return retval;
|
2005-07-04 20:26:53 +00:00
|
|
|
}
|
|
|
|
|
2006-08-15 17:23:48 +00:00
|
|
|
static struct commit_list *complete;
|
2005-10-19 21:27:02 +00:00
|
|
|
|
2006-09-21 05:02:01 +00:00
|
|
|
static int mark_complete(const char *path, const unsigned char *sha1, int flag, void *cb_data)
|
2005-10-19 21:27:02 +00:00
|
|
|
{
|
|
|
|
struct object *o = parse_object(sha1);
|
|
|
|
|
2006-07-12 03:45:31 +00:00
|
|
|
while (o && o->type == OBJ_TAG) {
|
2005-10-20 04:55:49 +00:00
|
|
|
struct tag *t = (struct tag *) o;
|
|
|
|
if (!t->tagged)
|
|
|
|
break; /* broken repository */
|
2005-10-19 21:27:02 +00:00
|
|
|
o->flags |= COMPLETE;
|
2005-10-20 04:55:49 +00:00
|
|
|
o = parse_object(t->tagged->sha1);
|
2005-10-19 21:27:02 +00:00
|
|
|
}
|
2006-07-12 03:45:31 +00:00
|
|
|
if (o && o->type == OBJ_COMMIT) {
|
2005-10-19 21:27:02 +00:00
|
|
|
struct commit *commit = (struct commit *)o;
|
|
|
|
commit->object.flags |= COMPLETE;
|
|
|
|
insert_by_date(commit, &complete);
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void mark_recent_complete_commits(unsigned long cutoff)
|
|
|
|
{
|
|
|
|
while (complete && cutoff <= complete->item->date) {
|
|
|
|
if (verbose)
|
|
|
|
fprintf(stderr, "Marking %s as complete\n",
|
|
|
|
sha1_to_hex(complete->item->object.sha1));
|
|
|
|
pop_most_recent_commit(&complete, COMPLETE);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-10-28 02:47:07 +00:00
|
|
|
static void filter_refs(struct ref **refs, int nr_match, char **match)
|
|
|
|
{
|
2006-05-11 22:28:44 +00:00
|
|
|
struct ref **return_refs;
|
|
|
|
struct ref *newlist = NULL;
|
|
|
|
struct ref **newtail = &newlist;
|
|
|
|
struct ref *ref, *next;
|
|
|
|
struct ref *fastarray[32];
|
|
|
|
|
|
|
|
if (nr_match && !fetch_all) {
|
|
|
|
if (ARRAY_SIZE(fastarray) < nr_match)
|
|
|
|
return_refs = xcalloc(nr_match, sizeof(struct ref *));
|
|
|
|
else {
|
|
|
|
return_refs = fastarray;
|
|
|
|
memset(return_refs, 0, sizeof(struct ref *) * nr_match);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
return_refs = NULL;
|
|
|
|
|
|
|
|
for (ref = *refs; ref; ref = next) {
|
|
|
|
next = ref->next;
|
|
|
|
if (!memcmp(ref->name, "refs/", 5) &&
|
|
|
|
check_ref_format(ref->name + 5))
|
|
|
|
; /* trash */
|
2006-11-24 15:00:13 +00:00
|
|
|
else if (fetch_all &&
|
Mechanical conversion to use prefixcmp()
This mechanically converts strncmp() to use prefixcmp(), but only when
the parameters match specific patterns, so that they can be verified
easily. Leftover from this will be fixed in a separate step, including
idiotic conversions like
if (!strncmp("foo", arg, 3))
=>
if (!(-prefixcmp(arg, "foo")))
This was done by using this script in px.perl
#!/usr/bin/perl -i.bak -p
if (/strncmp\(([^,]+), "([^\\"]*)", (\d+)\)/ && (length($2) == $3)) {
s|strncmp\(([^,]+), "([^\\"]*)", (\d+)\)|prefixcmp($1, "$2")|;
}
if (/strncmp\("([^\\"]*)", ([^,]+), (\d+)\)/ && (length($1) == $3)) {
s|strncmp\("([^\\"]*)", ([^,]+), (\d+)\)|(-prefixcmp($2, "$1"))|;
}
and running:
$ git grep -l strncmp -- '*.c' | xargs perl px.perl
Signed-off-by: Junio C Hamano <junkio@cox.net>
2007-02-20 09:53:29 +00:00
|
|
|
(!depth || prefixcmp(ref->name, "refs/tags/") )) {
|
2006-05-11 22:28:44 +00:00
|
|
|
*newtail = ref;
|
|
|
|
ref->next = NULL;
|
|
|
|
newtail = &ref->next;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
int order = path_match(ref->name, nr_match, match);
|
|
|
|
if (order) {
|
|
|
|
return_refs[order-1] = ref;
|
|
|
|
continue; /* we will link it later */
|
|
|
|
}
|
|
|
|
}
|
|
|
|
free(ref);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!fetch_all) {
|
|
|
|
int i;
|
|
|
|
for (i = 0; i < nr_match; i++) {
|
|
|
|
ref = return_refs[i];
|
|
|
|
if (ref) {
|
|
|
|
*newtail = ref;
|
|
|
|
ref->next = NULL;
|
|
|
|
newtail = &ref->next;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (return_refs != fastarray)
|
|
|
|
free(return_refs);
|
2005-10-28 02:47:07 +00:00
|
|
|
}
|
2006-05-11 22:28:44 +00:00
|
|
|
*refs = newlist;
|
2005-10-28 02:47:07 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int everything_local(struct ref **refs, int nr_match, char **match)
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
{
|
2005-10-19 21:27:02 +00:00
|
|
|
struct ref *ref;
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
int retval;
|
2005-10-19 21:27:02 +00:00
|
|
|
unsigned long cutoff = 0;
|
|
|
|
|
|
|
|
track_object_refs = 0;
|
|
|
|
save_commit_buffer = 0;
|
|
|
|
|
2005-10-28 02:47:07 +00:00
|
|
|
for (ref = *refs; ref; ref = ref->next) {
|
2005-10-19 21:27:02 +00:00
|
|
|
struct object *o;
|
|
|
|
|
|
|
|
o = parse_object(ref->old_sha1);
|
|
|
|
if (!o)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
/* We already have it -- which may mean that we were
|
|
|
|
* in sync with the other side at some time after
|
|
|
|
* that (it is OK if we guess wrong here).
|
|
|
|
*/
|
2006-07-12 03:45:31 +00:00
|
|
|
if (o->type == OBJ_COMMIT) {
|
2005-10-19 21:27:02 +00:00
|
|
|
struct commit *commit = (struct commit *)o;
|
|
|
|
if (!cutoff || cutoff < commit->date)
|
|
|
|
cutoff = commit->date;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-10-30 19:09:53 +00:00
|
|
|
if (!depth) {
|
|
|
|
for_each_ref(mark_complete, NULL);
|
|
|
|
if (cutoff)
|
|
|
|
mark_recent_complete_commits(cutoff);
|
|
|
|
}
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
|
2005-10-28 02:47:07 +00:00
|
|
|
/*
|
|
|
|
* Mark all complete remote refs as common refs.
|
|
|
|
* Don't mark them common yet; the server has to be told so first.
|
|
|
|
*/
|
|
|
|
for (ref = *refs; ref; ref = ref->next) {
|
2005-11-02 23:19:13 +00:00
|
|
|
struct object *o = deref_tag(lookup_object(ref->old_sha1),
|
|
|
|
NULL, 0);
|
2005-10-28 02:47:07 +00:00
|
|
|
|
2006-07-12 03:45:31 +00:00
|
|
|
if (!o || o->type != OBJ_COMMIT || !(o->flags & COMPLETE))
|
2005-10-28 02:47:07 +00:00
|
|
|
continue;
|
|
|
|
|
|
|
|
if (!(o->flags & SEEN)) {
|
|
|
|
rev_list_push((struct commit *)o, COMMON_REF | SEEN);
|
|
|
|
|
|
|
|
mark_common((struct commit *)o, 1, 1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
filter_refs(refs, nr_match, match);
|
|
|
|
|
|
|
|
for (retval = 1, ref = *refs; ref ; ref = ref->next) {
|
|
|
|
const unsigned char *remote = ref->old_sha1;
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
unsigned char local[20];
|
2005-10-19 21:27:02 +00:00
|
|
|
struct object *o;
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
|
2005-10-28 02:47:07 +00:00
|
|
|
o = lookup_object(remote);
|
2005-10-19 21:27:02 +00:00
|
|
|
if (!o || !(o->flags & COMPLETE)) {
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
retval = 0;
|
|
|
|
if (!verbose)
|
|
|
|
continue;
|
|
|
|
fprintf(stderr,
|
|
|
|
"want %s (%s)\n", sha1_to_hex(remote),
|
2005-10-28 02:47:07 +00:00
|
|
|
ref->name);
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2006-08-23 06:49:00 +00:00
|
|
|
hashcpy(ref->new_sha1, local);
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
if (!verbose)
|
|
|
|
continue;
|
|
|
|
fprintf(stderr,
|
|
|
|
"already have %s (%s)\n", sha1_to_hex(remote),
|
2005-10-28 02:47:07 +00:00
|
|
|
ref->name);
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
}
|
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
2006-11-01 22:06:23 +00:00
|
|
|
static pid_t setup_sideband(int fd[2], int xd[2])
|
|
|
|
{
|
|
|
|
pid_t side_pid;
|
|
|
|
|
|
|
|
if (!use_sideband) {
|
|
|
|
fd[0] = xd[0];
|
|
|
|
fd[1] = xd[1];
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
/* xd[] is talking with upload-pack; subprocess reads from
|
|
|
|
* xd[0], spits out band#2 to stderr, and feeds us band#1
|
|
|
|
* through our fd[0].
|
|
|
|
*/
|
|
|
|
if (pipe(fd) < 0)
|
|
|
|
die("fetch-pack: unable to set up pipe");
|
|
|
|
side_pid = fork();
|
|
|
|
if (side_pid < 0)
|
|
|
|
die("fetch-pack: unable to fork off sideband demultiplexer");
|
|
|
|
if (!side_pid) {
|
|
|
|
/* subprocess */
|
|
|
|
close(fd[0]);
|
|
|
|
if (xd[0] != xd[1])
|
|
|
|
close(xd[1]);
|
|
|
|
if (recv_sideband("fetch-pack", xd[0], fd[1], 2))
|
|
|
|
exit(1);
|
|
|
|
exit(0);
|
|
|
|
}
|
|
|
|
close(xd[0]);
|
|
|
|
close(fd[1]);
|
|
|
|
fd[1] = xd[1];
|
|
|
|
return side_pid;
|
|
|
|
}
|
|
|
|
|
2007-01-23 06:37:33 +00:00
|
|
|
static int get_pack(int xd[2])
|
2006-11-01 22:06:23 +00:00
|
|
|
{
|
|
|
|
int status;
|
|
|
|
pid_t pid, side_pid;
|
|
|
|
int fd[2];
|
2007-01-23 06:37:33 +00:00
|
|
|
const char *argv[20];
|
|
|
|
char keep_arg[256];
|
|
|
|
char hdr_arg[256];
|
|
|
|
const char **av;
|
|
|
|
int do_keep = keep_pack;
|
2006-11-01 22:06:23 +00:00
|
|
|
|
|
|
|
side_pid = setup_sideband(fd, xd);
|
2007-01-23 06:37:33 +00:00
|
|
|
|
|
|
|
av = argv;
|
|
|
|
*hdr_arg = 0;
|
2007-01-25 00:47:24 +00:00
|
|
|
if (unpack_limit) {
|
2007-01-23 06:37:33 +00:00
|
|
|
struct pack_header header;
|
|
|
|
|
|
|
|
if (read_pack_header(fd[0], &header))
|
|
|
|
die("protocol error: bad pack header");
|
|
|
|
snprintf(hdr_arg, sizeof(hdr_arg), "--pack_header=%u,%u",
|
|
|
|
ntohl(header.hdr_version), ntohl(header.hdr_entries));
|
2007-01-25 00:47:24 +00:00
|
|
|
if (ntohl(header.hdr_entries) < unpack_limit)
|
2007-01-23 06:37:33 +00:00
|
|
|
do_keep = 0;
|
|
|
|
else
|
|
|
|
do_keep = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (do_keep) {
|
|
|
|
*av++ = "index-pack";
|
|
|
|
*av++ = "--stdin";
|
|
|
|
if (!quiet)
|
|
|
|
*av++ = "-v";
|
|
|
|
if (use_thin_pack)
|
|
|
|
*av++ = "--fix-thin";
|
2007-01-25 00:47:24 +00:00
|
|
|
if (keep_pack > 1 || unpack_limit) {
|
2007-01-23 06:37:33 +00:00
|
|
|
int s = sprintf(keep_arg,
|
|
|
|
"--keep=fetch-pack %d on ", getpid());
|
|
|
|
if (gethostname(keep_arg + s, sizeof(keep_arg) - s))
|
|
|
|
strcpy(keep_arg + s, "localhost");
|
|
|
|
*av++ = keep_arg;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
*av++ = "unpack-objects";
|
|
|
|
if (quiet)
|
|
|
|
*av++ = "-q";
|
|
|
|
}
|
|
|
|
if (*hdr_arg)
|
|
|
|
*av++ = hdr_arg;
|
|
|
|
*av++ = NULL;
|
|
|
|
|
2006-11-01 22:06:23 +00:00
|
|
|
pid = fork();
|
|
|
|
if (pid < 0)
|
|
|
|
die("fetch-pack: unable to fork off %s", argv[0]);
|
|
|
|
if (!pid) {
|
|
|
|
dup2(fd[0], 0);
|
|
|
|
close(fd[0]);
|
|
|
|
close(fd[1]);
|
|
|
|
execv_git_cmd(argv);
|
|
|
|
die("%s exec failed", argv[0]);
|
|
|
|
}
|
|
|
|
close(fd[0]);
|
|
|
|
close(fd[1]);
|
|
|
|
while (waitpid(pid, &status, 0) < 0) {
|
|
|
|
if (errno != EINTR)
|
|
|
|
die("waiting for %s: %s", argv[0], strerror(errno));
|
|
|
|
}
|
|
|
|
if (WIFEXITED(status)) {
|
|
|
|
int code = WEXITSTATUS(status);
|
|
|
|
if (code)
|
|
|
|
die("%s died with error code %d", argv[0], code);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if (WIFSIGNALED(status)) {
|
|
|
|
int sig = WTERMSIG(status);
|
|
|
|
die("%s died of signal %d", argv[0], sig);
|
|
|
|
}
|
|
|
|
die("%s died of unnatural causes %d", argv[0], status);
|
|
|
|
}
|
|
|
|
|
2005-07-04 20:26:53 +00:00
|
|
|
static int fetch_pack(int fd[2], int nr_match, char **match)
|
|
|
|
{
|
2005-07-16 20:55:50 +00:00
|
|
|
struct ref *ref;
|
|
|
|
unsigned char sha1[20];
|
2005-07-04 20:26:53 +00:00
|
|
|
|
2005-10-28 02:47:07 +00:00
|
|
|
get_remote_heads(fd[0], &ref, 0, NULL, 0);
|
2006-10-30 19:09:06 +00:00
|
|
|
if (is_repository_shallow() && !server_supports("shallow"))
|
|
|
|
die("Server does not support shallow clients");
|
2005-10-28 02:50:26 +00:00
|
|
|
if (server_supports("multi_ack")) {
|
|
|
|
if (verbose)
|
|
|
|
fprintf(stderr, "Server supports multi_ack\n");
|
|
|
|
multi_ack = 1;
|
|
|
|
}
|
2006-09-10 23:27:08 +00:00
|
|
|
if (server_supports("side-band-64k")) {
|
|
|
|
if (verbose)
|
|
|
|
fprintf(stderr, "Server supports side-band-64k\n");
|
|
|
|
use_sideband = 2;
|
|
|
|
}
|
|
|
|
else if (server_supports("side-band")) {
|
2006-06-21 07:30:21 +00:00
|
|
|
if (verbose)
|
|
|
|
fprintf(stderr, "Server supports side-band\n");
|
|
|
|
use_sideband = 1;
|
|
|
|
}
|
2005-07-16 20:55:50 +00:00
|
|
|
if (!ref) {
|
|
|
|
packet_flush(fd[1]);
|
|
|
|
die("no matching remote head");
|
|
|
|
}
|
2005-10-28 02:47:07 +00:00
|
|
|
if (everything_local(&ref, nr_match, match)) {
|
git-fetch-pack: avoid unnecessary zero packing
If everything is up-to-date locally, we don't need to even ask for a
pack-file from the remote, or try to unpack it.
This is especially important for tags - since the pack-file common commit
logic is based purely on the commit history, it will never be able to find
a common tag, and will thus always end up re-fetching them.
Especially notably, if the tag points to a non-commit (eg a tagged tree),
the pack-file would be unnecessarily big, just because it cannot any most
recent common point between commits for pruning.
Short-circuiting the case where we already have that reference means that
we avoid a lot of these in the common case.
NOTE! This only matches remote ref names against the same local name,
which works well for tags, but is not as generic as it could be. If we
ever need to, we could match against _any_ local ref (if we have it, we
have it), but this "match against same name" is simpler and more
efficient, and covers the common case.
Renaming of refs is common for branch heads, but since those are always
commits, the pack-file generation can optimize that case.
In some cases we might still end up fetching pack-files unnecessarily, but
this at least avoids the re-fetching of tags over and over if you use a
regular
git fetch --tags ...
which was the main reason behind the change.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-10-18 18:35:17 +00:00
|
|
|
packet_flush(fd[1]);
|
|
|
|
goto all_done;
|
|
|
|
}
|
2005-08-12 09:08:29 +00:00
|
|
|
if (find_common(fd, sha1, ref) < 0)
|
2006-11-01 22:06:23 +00:00
|
|
|
if (keep_pack != 1)
|
2006-03-20 08:21:10 +00:00
|
|
|
/* When cloning, it is not unusual to have
|
|
|
|
* no common commit.
|
|
|
|
*/
|
|
|
|
fprintf(stderr, "warning: no common commits\n");
|
2005-12-15 06:17:38 +00:00
|
|
|
|
2007-01-23 06:37:33 +00:00
|
|
|
if (get_pack(fd))
|
2005-12-15 06:17:38 +00:00
|
|
|
die("git-fetch-pack: fetch failed.");
|
|
|
|
|
|
|
|
all_done:
|
|
|
|
while (ref) {
|
|
|
|
printf("%s %s\n",
|
|
|
|
sha1_to_hex(ref->old_sha1), ref->name);
|
|
|
|
ref = ref->next;
|
2005-07-04 23:35:13 +00:00
|
|
|
}
|
2005-12-15 06:17:38 +00:00
|
|
|
return 0;
|
2005-07-04 20:26:53 +00:00
|
|
|
}
|
|
|
|
|
2006-11-25 09:33:06 +00:00
|
|
|
static int remove_duplicates(int nr_heads, char **heads)
|
|
|
|
{
|
|
|
|
int src, dst;
|
|
|
|
|
|
|
|
for (src = dst = 0; src < nr_heads; src++) {
|
|
|
|
/* If heads[src] is different from any of
|
|
|
|
* heads[0..dst], push it in.
|
|
|
|
*/
|
|
|
|
int i;
|
|
|
|
for (i = 0; i < dst; i++) {
|
|
|
|
if (!strcmp(heads[i], heads[src]))
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (i < dst)
|
|
|
|
continue;
|
|
|
|
if (src != dst)
|
|
|
|
heads[dst] = heads[src];
|
|
|
|
dst++;
|
|
|
|
}
|
|
|
|
heads[dst] = 0;
|
|
|
|
return dst;
|
|
|
|
}
|
|
|
|
|
2007-01-25 00:47:24 +00:00
|
|
|
static int fetch_pack_config(const char *var, const char *value)
|
|
|
|
{
|
|
|
|
if (strcmp(var, "fetch.unpacklimit") == 0) {
|
2007-01-25 01:02:15 +00:00
|
|
|
fetch_unpack_limit = git_config_int(var, value);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (strcmp(var, "transfer.unpacklimit") == 0) {
|
|
|
|
transfer_unpack_limit = git_config_int(var, value);
|
2007-01-25 00:47:24 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return git_default_config(var, value);
|
|
|
|
}
|
|
|
|
|
2007-01-02 19:22:08 +00:00
|
|
|
static struct lock_file lock;
|
|
|
|
|
2005-07-04 20:26:53 +00:00
|
|
|
int main(int argc, char **argv)
|
|
|
|
{
|
|
|
|
int i, ret, nr_heads;
|
|
|
|
char *dest = NULL, **heads;
|
|
|
|
int fd[2];
|
|
|
|
pid_t pid;
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
struct stat st;
|
2005-07-04 20:26:53 +00:00
|
|
|
|
2005-11-26 08:47:59 +00:00
|
|
|
setup_git_directory();
|
2007-01-25 00:47:24 +00:00
|
|
|
git_config(fetch_pack_config);
|
2005-11-26 08:47:59 +00:00
|
|
|
|
2007-01-25 01:02:15 +00:00
|
|
|
if (0 <= transfer_unpack_limit)
|
|
|
|
unpack_limit = transfer_unpack_limit;
|
|
|
|
else if (0 <= fetch_unpack_limit)
|
|
|
|
unpack_limit = fetch_unpack_limit;
|
|
|
|
|
2005-07-04 20:26:53 +00:00
|
|
|
nr_heads = 0;
|
|
|
|
heads = NULL;
|
|
|
|
for (i = 1; i < argc; i++) {
|
|
|
|
char *arg = argv[i];
|
|
|
|
|
|
|
|
if (*arg == '-') {
|
Mechanical conversion to use prefixcmp()
This mechanically converts strncmp() to use prefixcmp(), but only when
the parameters match specific patterns, so that they can be verified
easily. Leftover from this will be fixed in a separate step, including
idiotic conversions like
if (!strncmp("foo", arg, 3))
=>
if (!(-prefixcmp(arg, "foo")))
This was done by using this script in px.perl
#!/usr/bin/perl -i.bak -p
if (/strncmp\(([^,]+), "([^\\"]*)", (\d+)\)/ && (length($2) == $3)) {
s|strncmp\(([^,]+), "([^\\"]*)", (\d+)\)|prefixcmp($1, "$2")|;
}
if (/strncmp\("([^\\"]*)", ([^,]+), (\d+)\)/ && (length($1) == $3)) {
s|strncmp\("([^\\"]*)", ([^,]+), (\d+)\)|(-prefixcmp($2, "$1"))|;
}
and running:
$ git grep -l strncmp -- '*.c' | xargs perl px.perl
Signed-off-by: Junio C Hamano <junkio@cox.net>
2007-02-20 09:53:29 +00:00
|
|
|
if (!(-prefixcmp(arg, "--upload-pack="))) {
|
2007-01-23 08:20:17 +00:00
|
|
|
uploadpack = arg + 14;
|
|
|
|
continue;
|
|
|
|
}
|
Mechanical conversion to use prefixcmp()
This mechanically converts strncmp() to use prefixcmp(), but only when
the parameters match specific patterns, so that they can be verified
easily. Leftover from this will be fixed in a separate step, including
idiotic conversions like
if (!strncmp("foo", arg, 3))
=>
if (!(-prefixcmp(arg, "foo")))
This was done by using this script in px.perl
#!/usr/bin/perl -i.bak -p
if (/strncmp\(([^,]+), "([^\\"]*)", (\d+)\)/ && (length($2) == $3)) {
s|strncmp\(([^,]+), "([^\\"]*)", (\d+)\)|prefixcmp($1, "$2")|;
}
if (/strncmp\("([^\\"]*)", ([^,]+), (\d+)\)/ && (length($1) == $3)) {
s|strncmp\("([^\\"]*)", ([^,]+), (\d+)\)|(-prefixcmp($2, "$1"))|;
}
and running:
$ git grep -l strncmp -- '*.c' | xargs perl px.perl
Signed-off-by: Junio C Hamano <junkio@cox.net>
2007-02-20 09:53:29 +00:00
|
|
|
if (!(-prefixcmp(arg, "--exec="))) {
|
2007-01-23 08:20:17 +00:00
|
|
|
uploadpack = arg + 7;
|
2005-07-14 07:08:37 +00:00
|
|
|
continue;
|
|
|
|
}
|
2005-12-18 09:55:29 +00:00
|
|
|
if (!strcmp("--quiet", arg) || !strcmp("-q", arg)) {
|
2005-08-12 09:08:29 +00:00
|
|
|
quiet = 1;
|
|
|
|
continue;
|
|
|
|
}
|
2005-12-18 09:55:29 +00:00
|
|
|
if (!strcmp("--keep", arg) || !strcmp("-k", arg)) {
|
2006-11-01 22:06:23 +00:00
|
|
|
keep_pack++;
|
2007-01-25 00:47:24 +00:00
|
|
|
unpack_limit = 0;
|
2007-01-23 06:37:33 +00:00
|
|
|
continue;
|
|
|
|
}
|
2006-02-20 08:38:39 +00:00
|
|
|
if (!strcmp("--thin", arg)) {
|
|
|
|
use_thin_pack = 1;
|
|
|
|
continue;
|
|
|
|
}
|
2006-03-20 08:21:10 +00:00
|
|
|
if (!strcmp("--all", arg)) {
|
|
|
|
fetch_all = 1;
|
|
|
|
continue;
|
|
|
|
}
|
2005-08-12 09:08:29 +00:00
|
|
|
if (!strcmp("-v", arg)) {
|
|
|
|
verbose = 1;
|
|
|
|
continue;
|
|
|
|
}
|
Mechanical conversion to use prefixcmp()
This mechanically converts strncmp() to use prefixcmp(), but only when
the parameters match specific patterns, so that they can be verified
easily. Leftover from this will be fixed in a separate step, including
idiotic conversions like
if (!strncmp("foo", arg, 3))
=>
if (!(-prefixcmp(arg, "foo")))
This was done by using this script in px.perl
#!/usr/bin/perl -i.bak -p
if (/strncmp\(([^,]+), "([^\\"]*)", (\d+)\)/ && (length($2) == $3)) {
s|strncmp\(([^,]+), "([^\\"]*)", (\d+)\)|prefixcmp($1, "$2")|;
}
if (/strncmp\("([^\\"]*)", ([^,]+), (\d+)\)/ && (length($1) == $3)) {
s|strncmp\("([^\\"]*)", ([^,]+), (\d+)\)|(-prefixcmp($2, "$1"))|;
}
and running:
$ git grep -l strncmp -- '*.c' | xargs perl px.perl
Signed-off-by: Junio C Hamano <junkio@cox.net>
2007-02-20 09:53:29 +00:00
|
|
|
if (!(-prefixcmp(arg, "--depth="))) {
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
depth = strtol(arg + 8, NULL, 0);
|
|
|
|
if (stat(git_path("shallow"), &st))
|
|
|
|
st.st_mtime = 0;
|
|
|
|
continue;
|
|
|
|
}
|
2005-07-04 20:26:53 +00:00
|
|
|
usage(fetch_pack_usage);
|
|
|
|
}
|
|
|
|
dest = arg;
|
|
|
|
heads = argv + i + 1;
|
|
|
|
nr_heads = argc - i - 1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (!dest)
|
|
|
|
usage(fetch_pack_usage);
|
2007-01-23 08:20:17 +00:00
|
|
|
pid = git_connect(fd, dest, uploadpack);
|
2005-07-04 20:26:53 +00:00
|
|
|
if (pid < 0)
|
|
|
|
return 1;
|
2006-11-25 09:33:06 +00:00
|
|
|
if (heads && nr_heads)
|
|
|
|
nr_heads = remove_duplicates(nr_heads, heads);
|
2005-07-04 20:26:53 +00:00
|
|
|
ret = fetch_pack(fd, nr_heads, heads);
|
|
|
|
close(fd[0]);
|
|
|
|
close(fd[1]);
|
2006-09-13 08:26:47 +00:00
|
|
|
ret |= finish_connect(pid);
|
2005-11-06 08:09:59 +00:00
|
|
|
|
|
|
|
if (!ret && nr_heads) {
|
|
|
|
/* If the heads to pull were given, we should have
|
|
|
|
* consumed all of them by matching the remote.
|
|
|
|
* Otherwise, 'git-fetch remote no-such-ref' would
|
|
|
|
* silently succeed without issuing an error.
|
|
|
|
*/
|
|
|
|
for (i = 0; i < nr_heads; i++)
|
|
|
|
if (heads[i] && heads[i][0]) {
|
|
|
|
error("no such remote ref %s", heads[i]);
|
|
|
|
ret = 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
if (!ret && depth > 0) {
|
|
|
|
struct cache_time mtime;
|
|
|
|
char *shallow = git_path("shallow");
|
|
|
|
int fd;
|
|
|
|
|
|
|
|
mtime.sec = st.st_mtime;
|
|
|
|
#ifdef USE_NSEC
|
|
|
|
mtime.usec = st.st_mtim.usec;
|
|
|
|
#endif
|
|
|
|
if (stat(shallow, &st)) {
|
|
|
|
if (mtime.sec)
|
|
|
|
die("shallow file was removed during fetch");
|
|
|
|
} else if (st.st_mtime != mtime.sec
|
|
|
|
#ifdef USE_NSEC
|
|
|
|
|| st.st_mtim.usec != mtime.usec
|
|
|
|
#endif
|
|
|
|
)
|
|
|
|
die("shallow file was changed during fetch");
|
|
|
|
|
|
|
|
fd = hold_lock_file_for_update(&lock, shallow, 1);
|
|
|
|
if (!write_shallow_commits(fd, 0)) {
|
2006-11-24 14:58:04 +00:00
|
|
|
unlink(shallow);
|
allow cloning a repository "shallowly"
By specifying a depth, you can now clone a repository such that
all fetched ancestor-chains' length is at most "depth". For example,
if the upstream repository has only 2 branches ("A" and "B"), which
are linear, and you specify depth 3, you will get A, A~1, A~2, A~3,
B, B~1, B~2, and B~3. The ends are automatically made shallow
commits.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-10-30 19:09:29 +00:00
|
|
|
rollback_lock_file(&lock);
|
|
|
|
} else {
|
|
|
|
close(fd);
|
|
|
|
commit_lock_file(&lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-09-13 08:26:47 +00:00
|
|
|
return !!ret;
|
2005-07-04 20:26:53 +00:00
|
|
|
}
|