If commands modify the git config too rapidly we might not notice
the .git/config file has been modified, as they could run in the
same filesystem timestamp window and thus not cause a change on
the config's mtime. This can cause repo to miss re-reading the
config file after running a command.
Allowing the cache to be flushed forces us to re-read the config.
Signed-off-by: Shawn O. Pearce <sop@google.com>
If a file (e.g. ~/.gitconfig) does not exist, we get None
here rather than a string. NoneType lacks rstrip() so we
cannot strip it.
Signed-off-by: Shawn O. Pearce <sop@google.com>
If there are shell special characters in the editor string, we must
use /bin/sh to parse and execute it, rather than trying to rely on
a simple split(' '). This avoids vim starting up with two empty
buffers, due to a misparsed command line.
Signed-off-by: Shawn O. Pearce <sop@google.com>
When someone copies and pastes a setup line from a web page,
they might actually copy 'repo sync' onto the clipboard and wind
up pasting it into the "Your Name" prompt. This means they will
initialize their client with the user name of "repo sync", creating
some rather funny looking commits later on. For example:
To setup your source tree:
mkdir ~/code
cd ~/code
repo init -u git://....
repo sync
If this entire block was just blindly copy and pasted into the
terminal, the shell won't read "repo sync" but "repo init" will.
By showing the user their full identity string, and asking them
to confirm it before we continue, we can give the hapless user a
chance to recover from this mistake, without unfairly harming those
who were actually named 'repo' by their parents.
Signed-off-by: Shawn O. Pearce <sop@google.com>
A git-config entry with no value was preventing repo
from initializing. This modifies _ReadGit() to handle
config entries with empty values.
Signed-off-by: David Aguilar <davvid@gmail.com>
Reported-by: Josh Guilfoyle <jasta00@gmail.com>
If the manifest repository is on a detached HEAD and we are parsing
an XML formatted manifest we should simply set the branch property
to None, rather than crash with an AttributeError.
Signed-off-by: Shawn O. Pearce <sop@google.com>
Rather than failing with no information, display the child exit
status and the command line we tried to use to edit a text file.
There may be some useful information to help understand the crash.
Signed-off-by: Shawn O. Pearce <sop@google.com>
If the SSH client terminated abnormally in the background (e.g. the
server shutdown while we were doing a sync) then the pid won't exist.
Instead of crashing, ignore it, the result we wanted (a non-orphaned
ssh process) is already acheived.
Signed-off-by: Shawn O. Pearce <sop@google.com>
The --color flag wasn't introduced until git 1.6.3. Prior to that
version, `git grep --color` just produces a fatal error, as it is
an unsupported option. Since this is just pretty output and is not
critical to execution, we can simply omit the option if the version
of git we are running on doesn't support it.
Signed-off-by: Shawn O. Pearce <sop@google.com>
This way we can use it to detect feature support in the underlying
git, such as new options or commands that have been added in more
recent versions.
Signed-off-by: Shawn O. Pearce <sop@google.com>
If the pickle config file is 0 bytes in length, we may have
crashed (or been aborted) while writing the file out to disk.
Instead of crashing with a backtrace, just treat the file as
though it wasn't present and load off a `git config` fork.
Signed-off-by: Shawn O. Pearce <sop@google.com>
Noticed by users on repo-discuss, we were missing a return False
here to signal that SSH control master was not used to setup the
network connection.
Signed-off-by: Shawn O. Pearce <sop@google.com>
If a line is blank in project.list, its not a relevant project path,
so skip over it. Existing project.list files may have blank lines if
sync was run with no projects at all, and the file was created empty.
Signed-off-by: Shawn O. Pearce <sop@google.com>
We have no working tree, so we cannot update the project.list
state file, nor should we try to delete a directory if a project is
removed from the manifest. Clients would still need the repository
for historical records.
Signed-off-by: Shawn O. Pearce <sop@google.com>
We accidentally introduced this message during 1.6.8 by always
invoking `git rebase` when there were no new commits from the
upstream, but the user had local commits.
Signed-off-by: Shawn O. Pearce <sop@google.com>
After a repo sync, some of the project paths might need
to be removed. This changes maintains a list of project
paths from the previous sync operation and compares.
The revisionExpr field now holds an expression from the manifest,
such as "refs/heads/master", while revisionId holds the current
commit-ish SHA-1 of the revisionExpr. Currently that is only
filled in if the manifest points directly to a SHA-1.
Signed-off-by: Shawn O. Pearce <sop@google.com>
The trick of looking at the reflog for the remote tracking branch
and only going back one commit works some of the time, but not all of
the time. Its sort of relying on the fact that the user didn't use
`repo sync -n` or `git fetch` to only update the tracking branches
and skip the working directory update.
Doing this right requires looking through the history of the SHA-1
source (what the upstream used to be) and finding a spot where the
DAG diveraged away suddenly, and consider that to be the rewind
point. That's really difficult to do, as we don't have a clear
picture of what that old point was.
A close approximation is to list all of the commits that are in
HEAD, but not the new upstream, and rebase all of those where the
committer email address is this user's email address. In most cases,
this will effectively rebase only the user's new original work.
If the user is the project maintainer and rewound the branch
themselves, and they don't want all of the commits they have created
to be rebased onto the new upstream, they should handle the rebase
on their own, after the sync is complete.
Signed-off-by: Shawn O. Pearce <sop@google.com>
We now feed Project a RemoteSpec, instead of the Remote directly
from the XmlManifest. This way the RemoteSpec already has the
full project URL, rather than just the base, permitting other
types of manifests to produce the URL in their own style.
Signed-off-by: Shawn O. Pearce <sop@google.com>
We'll soon be supporting two different manifest formats, but we
can't immediately remove support for the current XML one that is
in wide spread use within Android.
Signed-off-by: Shawn O. Pearce <sop@google.com>
This way we can put it in another directory than the config file
itself, e.g. hide it inside ".git" when parsing a ".gitmodules"
file from the working tree.
Signed-off-by: Shawn O. Pearce <sop@google.com>
This can be useful when pulling apart a configuration file, like
finding all entries which match submodule.*.*.
Signed-off-by: Shawn O. Pearce <sop@google.com>
These aren't that widely used, and actually make it difficult for
users to fully mirror a forest of repositories, and then permit
someone else to clone off that forest, rather then the original
upstream servers.
Signed-off-by: Shawn O. Pearce <sop@google.com>
We haven't supported this in a while, but the parser was still here.
Its all dead code, so strip it out.
Signed-off-by: Shawn O. Pearce <sop@google.com>
Extensive discussion with users lead to the fact that needing to
supply -a to view what they really wanted to see was just wrong.
Signed-off-by: Shawn O. Pearce <sop@google.com>
I only tested this with ssh://hostname/ style URLs, so I failed
to test ssh://user@hostname/ format, which failed if the hostname
portion was longer than 1 character.
Signed-off-by: Shawn O. Pearce <sop@google.com>
If the SSH URL doesn't contain a port number, but uses the ssh://
or git+ssh:// syntax we raised a Python runtime error due to the
'port' local variable not being assigned a value. Default it to
the IANA assigned port for SSH, 22.
Signed-off-by: Shawn O. Pearce <sop@google.com>
Performance improvements in repo sync caused us to skip out of the
initial Sync_LocalHalf without ever running CopyFiles, so we didn't
create the top level Makefile in new clients whose manifest request
one with a <copyfile> element.
Now we run CopyFiles after the initial read-tree that populates
the project working directory.
Signed-off-by: Shawn O. Pearce <sop@google.com>
If the current branch is published, but all published commits are
merged into the manifest revision, but there is also at least one
unpublished commit on the current branch, we should rebase the
unpublished commit, rather than creating a merge commit.
Signed-off-by: Shawn O. Pearce <sop@google.com>
The level 2 headings (denoted by ~) indent the heading two spaces,
but continue to use the bold formatter to offset them from the
other surrounding text.
Signed-off-by: Shawn O. Pearce <sop@google.com>
Mac OS X sets TMPDIR to a very long path within /var, so long
that a socket created in that location is too big for a struct
sockaddr_un on the platform, resulting in OpenSSH being unable
to create or bind to a socket in that location.
Instead we try to use the very short and very common /tmp, but
fall back to the guessed default if /tmp does not exist.
Signed-off-by: Shawn O. Pearce <sop@google.com>
By creating a background ssh "control master" process which lives
for the duration of our sync cycle we can easily cut the time for
a no-op sync of 132 projects from 60s to 18s.
Bug: REPO-11
Signed-off-by: Shawn O. Pearce <sop@google.com>
This way its clear the command did something, and reported
that it had nothing to show you, because you have no active
branches in this client.
Signed-off-by: Shawn O. Pearce <sop@google.com>
