Since f2690487 (fetch: opportunistically update tracking refs,
2013-05-11), we stopped taking a non-storing refspec given on the
command line of "git fetch" literally, and instead started mapping
it via remote.$name.fetch refspecs.  This allows

    $ git fetch origin master

from the 'origin' repository, which is configured with

    [remote "origin"]
        fetch = +refs/heads/*:refs/remotes/origin/*

to update refs/remotes/origin/master with the result, as if the
command line were

    $ git fetch origin +master:refs/remotes/origin/master

to reduce surprises and improve usability.  Before that change, a
refspec on the command line without a colon was only to fetch the
history and leave the result in FETCH_HEAD, without updating the
remote-tracking branches.

When you are simulating a fetch from you by your mothership with a
push by you into your mothership, instead of having:

    [remote "satellite"]
        fetch = +refs/heads/*:refs/remotes/satellite/*

on the mothership repository and running:

    mothership$ git fetch satellite

you would have:

    [remote "mothership"]
        push = +refs/heads/*:refs/remotes/satellite/*

on your satellite machine, and run:

    satellite$ git push mothership

Because we so far did not make the corresponding change to the push
side, this command:

    satellite$ git push mothership master

does _not_ allow you on the satellite to only push 'master' out but
still to the usual destination (i.e. refs/remotes/satellite/master).

Implement the logic to map an unqualified refspec given on the
command line via the remote.$name.push refspec.  This will bring a
bit more symmetry between "fetch" and "push".
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This teaches the deepest part of the callchain for "git push" (and
"git send-pack") to enforce "the old value of the ref must be this,
otherwise fail this push" (aka "compare-and-swap" / "--lockref").
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This plugs the push_cas_option data collected by the command line
option parser to the transport system with a new function
apply_push_cas(), which is called after match_push_refs() has
already been called.

At this point, we know which remote we are talking to, and what
remote refs we are going to update, so we can fill in the details
that may have been missing from the command line, such as

 (1) what abbreviated refname the user gave us matches the actual
     refname at the remote; and

 (2) which remote-tracking branch in our local repository to read
     the value of the object to expect at the remote.

to populate the old_sha1_expect[] field of each of the remote ref.
As stated in the documentation, the use of remote-tracking branch
as the default is a tentative one, and we may come up with a better
logic as we gain experience.

Still nobody uses this information, which is the topic of the next
patch.
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Update "git push" and "git send-pack" to parse this commnd line
option.

The intended sematics is:

 * "--force-with-lease" alone, without specifying the details, will
   protect _all_ remote refs that are going to be updated by
   requiring their current value to be the same as some reasonable
   default, unless otherwise specified;

 * "--force-with-lease=refname", without specifying the expected
   value, will protect that refname, if it is going to be updated,
   by requiring its current value to be the same as some reasonable
   default.

 * "--force-with-lease=refname:value" will protect that refname, if
   it is going to be updated, by requiring its current value to be
   the same as the specified value; and

 * "--no-force-with-lease" will cancel all the previous --force-with-lease on the
   command line.

For now, "some reasonable default" is tentatively defined as "the
value of the remote-tracking branch we have for the ref of the
remote being updated", and it is an error if we do not have such a
remote-tracking branch.  But this is known to be fragile, its use is
not yet recommended, and hopefully we will find more reasonable
default as we gain experience with this feature.  The manual marks
the feature as experimental unless the expected value is specified
explicitly for this reason.

Because the command line options are parsed _before_ we know which
remote we are pushing to, there needs further processing to the
parsed data after we instantiate the transport object to:

 * expand "refname" given by the user to a full refname to be
   matched with the list of "struct ref" used in match_push_refs()
   and set_ref_status_for_push(); and

 * learning the actual local ref that is the remote-tracking branch
   for the specified remote ref.

Further, some processing need to be deferred until we find the set
of remote refs and match_push_refs() returns in order to find the
ones that need to be checked after explicit ones have been processed
for "--force-with-lease" (no specific details).

These post-processing will be the topic of the next patch.

This option was originally called "cas" (for "compare and swap"),
the name which nobody liked because it was too technical.  The
second attempt called it "lockref" (because it is conceptually like
pushing after taking a lock) but the word "lock" was hated because
it implied that it may reject push by others, which is not the way
this option works.  This round calls it "force-with-lease".  You
assume you took the lease on the ref when you fetched to decide what
the rebased history should be, and you can push back only if the
lease has not been broken.
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Without "git fetch --prune", remote-tracking branches for a branch
the other side already has removed will stay forever.  Some people
want to always run "git fetch --prune".

To accommodate users who want to either prune always or when fetching
from a particular remote, add two new configuration variables
"fetch.prune" and "remote.<name>.prune":

 - "fetch.prune" allows to enable prune for all fetch operations.

 - "remote.<name>.prune" allows to change the behaviour per remote.

The latter will naturally override the former, and the --[no-]prune
option from the command line will override the configured default.

Since --prune is a potentially destructive operation (Git doesn't
keep reflogs for deleted references yet), we don't want to prune
without users consent, so this configuration will not be on by
default.
Helped-by: NJunio C Hamano <gitster@pobox.com>
Signed-off-by: NMichael Schubert <mschub@elegosoft.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

The definition of "struct ref" in "cache.h", a header file so
central to the system, always confused me.  This structure is not
about the local ref used by sha1-name API to name local objects.

It is what refspecs are expanded into, after finding out what refs
the other side has, to define what refs are updated after object
transfer succeeds to what values.  It belongs to "remote.h" together
with "struct refspec".

While we are at it, also move the types and functions related to the
Git transport connection to a new header file connect.h
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Currently, do_push() in push.c calls remote_get(), which gets the
configured remote for fetching and pushing.  Replace this call with a
call to pushremote_get() instead, a new function that will return the
remote configured specifically for pushing.  This function tries to
work with the string pushremote_name, before falling back to the
codepath of remote_get().  This patch has no visible impact, but
serves to enable future patches to introduce configuration variables
to set pushremote_name.  For example, you can now do the following in
handle_config():

    if (!strcmp(key, "remote.pushdefault"))
       git_config_string(&pushremote_name, key, value);

Then, pushes will automatically go to the remote specified by
remote.pushdefault.
Signed-off-by: NRamkumar Ramachandra <artagnon@gmail.com>
Reviewed-by: NJeff King <peff@peff.net>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

The new option "--follow-tags" tells "git push" to push annotated
tags that are missing from the other side and that can be reached by
the history that is otherwise pushed out.

For example, if you are using the "simple", "current", or "upstream"
push, you would ordinarily push the history leading to the commit at
your current HEAD and nothing else.  With this option, you would
also push all annotated tags that can be reached from that commit to
the other side.
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Teach "git fetch" to accept an exact SHA-1 object name the user may
obtain out of band on the LHS of a pathspec, and send it on a "want"
message when the server side advertises the allow-tip-sha1-in-want
capability.
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Since we store lists of refs as linked lists, we can use
llist_mergesort to efficiently sort them.
Signed-off-by: NJeff King <peff@peff.net>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

When pushing groups of refs to a remote, there is no simple way to remove
old refs that still exist at the remote that is no longer updated from us.
This will allow us to remove such refs from the remote.

With this change, running this command

 $ git push --prune remote refs/heads/*:refs/remotes/laptop/*

removes refs/remotes/laptop/foo from the remote if we do not have branch
"foo" locally anymore.
Signed-off-by: NFelipe Contreras <felipe.contreras@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

If the user gave us refspecs on the command line, we should use those
when deciding whether to prune a ref instead of relying on the
refspecs in the config.

Previously, running

    git fetch --prune origin refs/heads/master:refs/remotes/origin/master

would delete every other ref under the origin namespace because we
were using the refspec to filter the available refs but using the
configured refspec to figure out if a ref had been deleted on the
remote. This is clearly the wrong thing to do.

Change prune_refs and get_stale_heads to simply accept a list of
references and a list of refspecs. The caller of either function needs
to decide what refspecs should be used to decide whether a ref is
stale.
Signed-off-by: NCarlos Martín Nieto <cmn@elego.de>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Yes, there is a warning that says the function is only used by push in big
red letters in front of this function, but it didn't say a more important
thing it should have said: what the function is for and what it does.

Rename it and document it to avoid future confusion.
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This is a useful function, and we have already made the
similar alloc_ref and copy_ref_list available.
Signed-off-by: NJeff King <peff@peff.net>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Without this patch at least IBM VisualAge C 5.0 (I have 5.0.2) on AIX
5.1 fails to compile git.

enum style is inconsistent already, with some enums declared on one
line, some over 3 lines with the enum values all on the middle line,
sometimes with 1 enum value per line... and independently of that the
trailing comma is sometimes present and other times absent, often
mixing with/without trailing comma styles in a single file, and
sometimes in consecutive enum declarations.

Clearly, omitting the comma is the more portable style, and this patch
changes all enum declarations to use the portable omitted dangling
comma style consistently.
Signed-off-by: NGary V. Vaughan <gary@thewrittenword.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Move the logic that detects up-to-date and non-fast-forward refs to a
new function in remote.[ch], set_ref_status_for_push().

Make transport_push() invoke set_ref_status_for_push() before invoking
the push_refs() implementation. (As a side-effect, the push_refs()
implementation in transport-helper.c now knows of non-fast-forward
pushes.)

Removed logic for detecting up-to-date refs from the push_refs()
implementation in transport-helper.c, as transport_push() has already
done so for it.

Make cmd_send_pack() invoke set_ref_status_for_push() before invoking
send_pack(), as transport_push() can't do it for send_pack() here.

Mark the test on the return status of non-fast-forward push to fail.
Git now exits with success, as transport.c::transport_push() does not
check for refs with status REF_STATUS_REJECT_NONFASTFORWARD nor does it
indicate rejected pushes with its return value.

Mark the test for ref status to succeed. As mentioned earlier, refs
might be marked as non-fast-forwards, triggering the push status
printing mechanism in transport.c.
Signed-off-by: NTay Ray Chuan <rctay89@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This allows a helper to say that, when it handles "import
refs/heads/topic", the script it outputs will actually write to
refs/svn/origin/branches/topic; therefore, transport-helper should
read it from the latter location after git-fast-import completes.
Signed-off-by: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

If this is set, the url is not required, and the transport always uses
a helper named "git-remote-<value>".

It is a separate configuration option in order to allow a sensible
configuration for foreign systems which either have no meaningful urls
for repositories or which require urls that do not specify the system
used by the repository at that location. However, this only affects
how the name of the helper is determined, not anything about the
interaction with the helper, and the contruction is such that, if the
foreign scm does happen to use a co-named url method, a url with that
method may be used directly.
Signed-off-by: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NSverre Rabbelier <srabbelier@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Move the logic in builtin-remote.c which determines which local heads are stale
to remote.c so it can be used by other builtins.
Signed-off-by: NJay Soffian <jaysoffian@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This introduces a config setting remote.$remotename.pushurl which is
used for pushes only. If absent remote.$remotename.url is used for
pushes and fetches as before.
This is useful, for example, in order to do passwordless fetches
(remote update) over the git transport but pushes over ssh.
Signed-off-by: NMichael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Avoid code duplication by moving list tail search to match_refs().

This does not change the semantics, except for http-push, which now inserts
to the front of the ref list in order to get rid of the global remote_tail.
Signed-off-by: NClemens Buchacher <drizzd@aon.at>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Previously, there was no easy way to check for the existence of a
configured remote. remote_get for example would always create the remote
"on demand".

This new function returns 1 if the remote is configured, 0 otherwise.
Signed-off-by: NFinn Arne Gangstad <finnag@pvv.org>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Determining HEAD is ambiguous since it is done by comparing SHA1s.

In the case of multiple matches we return refs/heads/master if it
matches, else we return the first match we encounter. builtin-remote
needs all matches returned to it, so add a flag for it to request such.

To be simple and consistent, the return value is now a copy (including
peer_ref) of the matching refs.

Originally contributed by Jeff King along with the prior commit as a
single patch.
Signed-off-by: NJay Soffian <jaysoffian@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This function had complications which made it hard to extend.

- It used to do two things: find the HEAD ref, and then find a
  matching ref, optionally returning the former via assignment to a
  passed-in pointer. Since finding HEAD is a one-liner, just have a
  caller do it themselves and pass it as an argument.

- It used to manually search through the ref list for
  refs/heads/master; this can be a one-line call to
  find_ref_by_name.

Originally contributed by Jeff King along with the next commit as a
single patch.
Signed-off-by: NJay Soffian <jaysoffian@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Move locate_head() to remote.c and rename it to guess_remote_head() to
more accurately reflect what it does. This is in preparation for being
able to call it from builtin-remote.c
Signed-off-by: NJay Soffian <jaysoffian@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

ref_newer() appears to have been copied from builtin-send-pack.c to
http-push.c via cut and paste. This patch moves the function and its
helper unmark_and_free() to remote.c. There was a slight difference
between the two implementations, one used TMP_MARK for the mark, the
other used 1. Per Jeff King, I went with TMP_MARK as more correct.

This is in preparation for being able to call it from builtin-remote.c
Signed-off-by: NJay Soffian <jaysoffian@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

get_local_heads() appears to have been copied from builtin-send-pack.c
to http-push.c via cut and paste. This patch moves the function and its
helper one_local_ref() to remote.c.

The two copies of one_local_ref() were not identical. I used the more
recent version from builtin-send-pack.c after confirming with Jeff King
that it was an oversight that commit 30affa1e did not update both
copies.

This is in preparation for being able to call it from builtin-remote.c
Signed-off-by: NJay Soffian <jaysoffian@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This allows one to track where was the remote's original source, so that
it's possible to decide if it makes sense to migrate it to the config
format or not.
Signed-off-by: NMiklos Vajna <vmiklos@frugalware.org>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

With all calls to alloc_ref() gone, we can remove it and then we're free
to give alloc_ref_from_str() the shorter name.  It's a much nicer
interface, as the callers always need to have a name string when they
allocate a ref anyway and don't need to calculate and pass its length+1
any more.
Signed-off-by: NRene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

These functions are not used by any other file.
Signed-off-by: NNanako Shiraishi <nanako3@lavabit.com>
Signed-off-by: NShawn O. Pearce <spearce@spearce.org>

A number of call sites allocate memory for a refspec array, populate
its members with heap memory, and then free only the refspec pointer
while leaking the memory allocated for the member elements. Provide
a function for freeing the elements of a refspec array and the array
itself.

Caution to callers: code paths must be checked to ensure that the
refspec members "src" and "dst" can be passed to free.
Signed-off-by: NBrandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

People seem to like "Your branch is ahead by N commit" report made by
"git checkout", but the interface into the statistics function was a bit
clunky.  This splits the function into three parts:

 * The core "commit counting" function that takes "struct branch" and
   returns number of commits to show if we are ahead, behind or forked;

 * Convenience "stat formating" function that takes "struct branch" and
   formats the report into a given strbuf, using the above function;

 * "checkout" specific function that takes "branch_info" (type that is
   internal to checkout implementation), calls the above function and
   print the formatted result.

in the hope that the former two can be more easily reusable.
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

Also fix an underallocation in walker.c::interpret_target().
Signed-off-by: NKrzysztof Kowalczyk <kkowalczyk@gmail.com>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

The refspec refs/tags/*:refs/tags/* is sufficiently common and generic
to merit having a constant instead of generating it as needed.
Signed-off-by: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This patch provides a way to specify "push matching heads" using a
special refspec ":".  This is useful because it allows "push = +:"
as a way to specify that matching refs will be pushed but, in addition,
forced updates will be allowed, which was not possible before.
Signed-off-by: NPaolo Bonzini <bonzini@gnu.org>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This makes a struct ref able to represent a symref, and makes http.c
able to recognize one, and makes transport.c look for "HEAD" as a ref
in the list, and makes it dereference symrefs for the resulting ref,
if any.
Signed-off-by: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This patch adds a remote.*.mirror configuration option that,
when set, automatically puts git-push in --mirror mode for that
remote.

Furthermore, the option is set automatically by `git remote
add --mirror'.

The code in remote.c to parse remote.*.skipdefaultupdate
had a subtle problem: a comment in the code indicated that
special care was needed for boolean options, but this care was
not used in parsing the option.  Since I was touching related
code, I did this fix too.

[jc: and I further fixed up the "ignore boolean" code.]
Signed-off-by: NPaolo Bonzini <bonzini@gnu.org>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This can happen if the arguments to git-remote add is switched by the
user, and git would only show an error if fetching was also requested.
Fix it by using the refspec parsing engine to check if the requested
name can be parsed as a remote before add it.

Also cleanup so that the "remote.<name>.url" config name buffer is only
initialized once.
Signed-off-by: NJonas Fonseca <fonseca@diku.dk>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

This changes the pattern matching code to not store the required final
/ before the *, and then to require each side to be a valid ref (or
empty). In particular, any refspec that looks like it should be a
pattern but doesn't quite meet the requirements will be found to be
invalid as a fallback non-pattern.

This was cherry picked from commit ef00d150 (Tighten refspec processing,
2008-03-17), and two fix-up commits 46220ca1 (remote.c: Fix overtight
refspec validation, 2008-03-20) and 7d19da46 (refspec: allow colon-less
wildcard "refs/category/*", 2008-03-25) squashed in.
Signed-off-by: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NJunio C Hamano <gitster@pobox.com>

We tightened the refspec validation code in an earlier commit ef00d150
(Tighten refspec processing, 2008-03-17) per my suggestion, but the
suggestion was misguided to begin with and it broke this usage:

    $ git push origin HEAD~12:master

The syntax of push refspecs and fetch refspecs are similar in that they
are both colon separated LHS and RHS (possibly prefixed with a + to
force), but the similarity ends there.  For example, LHS in a push refspec
can be anything that evaluates to a valid object name at runtime (except
when colon and RHS is missing, or it is a glob), while it must be a
valid-looking refname in a fetch refspec.  To validate them correctly, the
caller needs to be able to say which kind of refspecs they are.  It is
unreasonable to keep a single interface that cannot tell which kind it is
dealing with, and ask it to behave sensibly.

This commit separates the parsing of the two into different functions, and
clarifies the code to implement the parsing proper (i.e. splitting into
two parts, making sure both sides are wildcard or neither side is).

This happens to also allow pushing a commit named with the esoteric "look
for that string" syntax:

    $ git push ../test.git ':/remote.c: Fix overtight refspec:master'
Signed-off-by: NJunio C Hamano <gitster@pobox.com>