Revision 2501aff8b7516115c409cb34cc50305cdde40a47 authored by Jeff King on 28 September 2013, 08:31:45 UTC, committed by Jonathan Nieder on 14 October 2013, 23:55:13 UTC
When we are handling a curl response code in http_request or in the remote-curl RPC code, we use the handle_curl_result helper to translate curl's response into an easy-to-use code. When we see an HTTP 401, we do one of two things: 1. If we already had a filled-in credential, we mark it as rejected, and then return HTTP_NOAUTH to indicate to the caller that we failed. 2. If we didn't, then we ask for a new credential and tell the caller HTTP_REAUTH to indicate that they may want to try again. Rejecting in the first case makes sense; it is the natural result of the request we just made. However, prompting for more credentials in the second step does not always make sense. We do not know for sure that the caller is going to make a second request, and nor are we sure that it will be to the same URL. Logically, the prompt belongs not to the request we just finished, but to the request we are (maybe) about to make. In practice, it is very hard to trigger any bad behavior. Currently, if we make a second request, it will always be to the same URL (even in the face of redirects, because curl handles the redirects internally). And we almost always retry on HTTP_REAUTH these days. The one exception is if we are streaming a large RPC request to the server (e.g., a pushed packfile), in which case we cannot restart. It's extremely unlikely to see a 401 response at this stage, though, as we would typically have seen it when we sent a probe request, before streaming the data. This patch drops the automatic prompt out of case 2, and instead requires the caller to do it. This is a few extra lines of code, and the bug it fixes is unlikely to come up in practice. But it is conceptually cleaner, and paves the way for better handling of credentials across redirects. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
1 parent 1bbcc22
git-p4.txt
git-p4(1)
=========
NAME
----
git-p4 - Import from and submit to Perforce repositories
SYNOPSIS
--------
[verse]
'git p4 clone' [<sync options>] [<clone options>] <p4 depot path>...
'git p4 sync' [<sync options>] [<p4 depot path>...]
'git p4 rebase'
'git p4 submit' [<submit options>] [<master branch name>]
DESCRIPTION
-----------
This command provides a way to interact with p4 repositories
using Git.
Create a new Git repository from an existing p4 repository using
'git p4 clone', giving it one or more p4 depot paths. Incorporate
new commits from p4 changes with 'git p4 sync'. The 'sync' command
is also used to include new branches from other p4 depot paths.
Submit Git changes back to p4 using 'git p4 submit'. The command
'git p4 rebase' does a sync plus rebases the current branch onto
the updated p4 remote branch.
EXAMPLE
-------
* Clone a repository:
+
------------
$ git p4 clone //depot/path/project
------------
* Do some work in the newly created Git repository:
+
------------
$ cd project
$ vi foo.h
$ git commit -a -m "edited foo.h"
------------
* Update the Git repository with recent changes from p4, rebasing your
work on top:
+
------------
$ git p4 rebase
------------
* Submit your commits back to p4:
+
------------
$ git p4 submit
------------
COMMANDS
--------
Clone
~~~~~
Generally, 'git p4 clone' is used to create a new Git directory
from an existing p4 repository:
------------
$ git p4 clone //depot/path/project
------------
This:
1. Creates an empty Git repository in a subdirectory called 'project'.
+
2. Imports the full contents of the head revision from the given p4
depot path into a single commit in the Git branch 'refs/remotes/p4/master'.
+
3. Creates a local branch, 'master' from this remote and checks it out.
To reproduce the entire p4 history in Git, use the '@all' modifier on
the depot path:
------------
$ git p4 clone //depot/path/project@all
------------
Sync
~~~~
As development continues in the p4 repository, those changes can
be included in the Git repository using:
------------
$ git p4 sync
------------
This command finds new changes in p4 and imports them as Git commits.
P4 repositories can be added to an existing Git repository using
'git p4 sync' too:
------------
$ mkdir repo-git
$ cd repo-git
$ git init
$ git p4 sync //path/in/your/perforce/depot
------------
This imports the specified depot into
'refs/remotes/p4/master' in an existing Git repository. The
'--branch' option can be used to specify a different branch to
be used for the p4 content.
If a Git repository includes branches 'refs/remotes/origin/p4', these
will be fetched and consulted first during a 'git p4 sync'. Since
importing directly from p4 is considerably slower than pulling changes
from a Git remote, this can be useful in a multi-developer environment.
If there are multiple branches, doing 'git p4 sync' will automatically
use the "BRANCH DETECTION" algorithm to try to partition new changes
into the right branch. This can be overridden with the '--branch'
option to specify just a single branch to update.
Rebase
~~~~~~
A common working pattern is to fetch the latest changes from the p4 depot
and merge them with local uncommitted changes. Often, the p4 repository
is the ultimate location for all code, thus a rebase workflow makes
sense. This command does 'git p4 sync' followed by 'git rebase' to move
local commits on top of updated p4 changes.
------------
$ git p4 rebase
------------
Submit
~~~~~~
Submitting changes from a Git repository back to the p4 repository
requires a separate p4 client workspace. This should be specified
using the 'P4CLIENT' environment variable or the Git configuration
variable 'git-p4.client'. The p4 client must exist, but the client root
will be created and populated if it does not already exist.
To submit all changes that are in the current Git branch but not in
the 'p4/master' branch, use:
------------
$ git p4 submit
------------
To specify a branch other than the current one, use:
------------
$ git p4 submit topicbranch
------------
The upstream reference is generally 'refs/remotes/p4/master', but can
be overridden using the '--origin=' command-line option.
The p4 changes will be created as the user invoking 'git p4 submit'. The
'--preserve-user' option will cause ownership to be modified
according to the author of the Git commit. This option requires admin
privileges in p4, which can be granted using 'p4 protect'.
OPTIONS
-------
General options
~~~~~~~~~~~~~~~
All commands except clone accept these options.
--git-dir <dir>::
Set the 'GIT_DIR' environment variable. See linkgit:git[1].
--verbose, -v::
Provide more progress information.
Sync options
~~~~~~~~~~~~
These options can be used in the initial 'clone' as well as in
subsequent 'sync' operations.
--branch <ref>::
Import changes into <ref> instead of refs/remotes/p4/master.
If <ref> starts with refs/, it is used as is. Otherwise, if
it does not start with p4/, that prefix is added.
+
By default a <ref> not starting with refs/ is treated as the
name of a remote-tracking branch (under refs/remotes/). This
behavior can be modified using the --import-local option.
+
The default <ref> is "master".
+
This example imports a new remote "p4/proj2" into an existing
Git repository:
+
----
$ git init
$ git p4 sync --branch=refs/remotes/p4/proj2 //depot/proj2
----
--detect-branches::
Use the branch detection algorithm to find new paths in p4. It is
documented below in "BRANCH DETECTION".
--changesfile <file>::
Import exactly the p4 change numbers listed in 'file', one per
line. Normally, 'git p4' inspects the current p4 repository
state and detects the changes it should import.
--silent::
Do not print any progress information.
--detect-labels::
Query p4 for labels associated with the depot paths, and add
them as tags in Git. Limited usefulness as only imports labels
associated with new changelists. Deprecated.
--import-labels::
Import labels from p4 into Git.
--import-local::
By default, p4 branches are stored in 'refs/remotes/p4/',
where they will be treated as remote-tracking branches by
linkgit:git-branch[1] and other commands. This option instead
puts p4 branches in 'refs/heads/p4/'. Note that future
sync operations must specify '--import-local' as well so that
they can find the p4 branches in refs/heads.
--max-changes <n>::
Limit the number of imported changes to 'n'. Useful to
limit the amount of history when using the '@all' p4 revision
specifier.
--keep-path::
The mapping of file names from the p4 depot path to Git, by
default, involves removing the entire depot path. With this
option, the full p4 depot path is retained in Git. For example,
path '//depot/main/foo/bar.c', when imported from
'//depot/main/', becomes 'foo/bar.c'. With '--keep-path', the
Git path is instead 'depot/main/foo/bar.c'.
--use-client-spec::
Use a client spec to find the list of interesting files in p4.
See the "CLIENT SPEC" section below.
Clone options
~~~~~~~~~~~~~
These options can be used in an initial 'clone', along with the 'sync'
options described above.
--destination <directory>::
Where to create the Git repository. If not provided, the last
component in the p4 depot path is used to create a new
directory.
--bare::
Perform a bare clone. See linkgit:git-clone[1].
-/ <path>::
Exclude selected depot paths when cloning.
Submit options
~~~~~~~~~~~~~~
These options can be used to modify 'git p4 submit' behavior.
--origin <commit>::
Upstream location from which commits are identified to submit to
p4. By default, this is the most recent p4 commit reachable
from 'HEAD'.
-M::
Detect renames. See linkgit:git-diff[1]. Renames will be
represented in p4 using explicit 'move' operations. There
is no corresponding option to detect copies, but there are
variables for both moves and copies.
--preserve-user::
Re-author p4 changes before submitting to p4. This option
requires p4 admin privileges.
--export-labels::
Export tags from Git as p4 labels. Tags found in Git are applied
to the perforce working directory.
--dry-run, -n::
Show just what commits would be submitted to p4; do not change
state in Git or p4.
--prepare-p4-only::
Apply a commit to the p4 workspace, opening, adding and deleting
files in p4 as for a normal submit operation. Do not issue the
final "p4 submit", but instead print a message about how to
submit manually or revert. This option always stops after the
first (oldest) commit. Git tags are not exported to p4.
--conflict=(ask|skip|quit)::
Conflicts can occur when applying a commit to p4. When this
happens, the default behavior ("ask") is to prompt whether to
skip this commit and continue, or quit. This option can be used
to bypass the prompt, causing conflicting commits to be automatically
skipped, or to quit trying to apply commits, without prompting.
--branch <branch>::
After submitting, sync this named branch instead of the default
p4/master. See the "Sync options" section above for more
information.
Rebase options
~~~~~~~~~~~~~~
These options can be used to modify 'git p4 rebase' behavior.
--import-labels::
Import p4 labels.
DEPOT PATH SYNTAX
-----------------
The p4 depot path argument to 'git p4 sync' and 'git p4 clone' can
be one or more space-separated p4 depot paths, with an optional
p4 revision specifier on the end:
"//depot/my/project"::
Import one commit with all files in the '#head' change under that tree.
"//depot/my/project@all"::
Import one commit for each change in the history of that depot path.
"//depot/my/project@1,6"::
Import only changes 1 through 6.
"//depot/proj1@all //depot/proj2@all"::
Import all changes from both named depot paths into a single
repository. Only files below these directories are included.
There is not a subdirectory in Git for each "proj1" and "proj2".
You must use the '--destination' option when specifying more
than one depot path. The revision specifier must be specified
identically on each depot path. If there are files in the
depot paths with the same name, the path with the most recently
updated version of the file is the one that appears in Git.
See 'p4 help revisions' for the full syntax of p4 revision specifiers.
CLIENT SPEC
-----------
The p4 client specification is maintained with the 'p4 client' command
and contains among other fields, a View that specifies how the depot
is mapped into the client repository. The 'clone' and 'sync' commands
can consult the client spec when given the '--use-client-spec' option or
when the useClientSpec variable is true. After 'git p4 clone', the
useClientSpec variable is automatically set in the repository
configuration file. This allows future 'git p4 submit' commands to
work properly; the submit command looks only at the variable and does
not have a command-line option.
The full syntax for a p4 view is documented in 'p4 help views'. 'git p4'
knows only a subset of the view syntax. It understands multi-line
mappings, overlays with '+', exclusions with '-' and double-quotes
around whitespace. Of the possible wildcards, 'git p4' only handles
'...', and only when it is at the end of the path. 'git p4' will complain
if it encounters an unhandled wildcard.
Bugs in the implementation of overlap mappings exist. If multiple depot
paths map through overlays to the same location in the repository,
'git p4' can choose the wrong one. This is hard to solve without
dedicating a client spec just for 'git p4'.
The name of the client can be given to 'git p4' in multiple ways. The
variable 'git-p4.client' takes precedence if it exists. Otherwise,
normal p4 mechanisms of determining the client are used: environment
variable P4CLIENT, a file referenced by P4CONFIG, or the local host name.
BRANCH DETECTION
----------------
P4 does not have the same concept of a branch as Git. Instead,
p4 organizes its content as a directory tree, where by convention
different logical branches are in different locations in the tree.
The 'p4 branch' command is used to maintain mappings between
different areas in the tree, and indicate related content. 'git p4'
can use these mappings to determine branch relationships.
If you have a repository where all the branches of interest exist as
subdirectories of a single depot path, you can use '--detect-branches'
when cloning or syncing to have 'git p4' automatically find
subdirectories in p4, and to generate these as branches in Git.
For example, if the P4 repository structure is:
----
//depot/main/...
//depot/branch1/...
----
And "p4 branch -o branch1" shows a View line that looks like:
----
//depot/main/... //depot/branch1/...
----
Then this 'git p4 clone' command:
----
git p4 clone --detect-branches //depot@all
----
produces a separate branch in 'refs/remotes/p4/' for //depot/main,
called 'master', and one for //depot/branch1 called 'depot/branch1'.
However, it is not necessary to create branches in p4 to be able to use
them like branches. Because it is difficult to infer branch
relationships automatically, a Git configuration setting
'git-p4.branchList' can be used to explicitly identify branch
relationships. It is a list of "source:destination" pairs, like a
simple p4 branch specification, where the "source" and "destination" are
the path elements in the p4 repository. The example above relied on the
presence of the p4 branch. Without p4 branches, the same result will
occur with:
----
git init depot
cd depot
git config git-p4.branchList main:branch1
git p4 clone --detect-branches //depot@all .
----
PERFORMANCE
-----------
The fast-import mechanism used by 'git p4' creates one pack file for
each invocation of 'git p4 sync'. Normally, Git garbage compression
(linkgit:git-gc[1]) automatically compresses these to fewer pack files,
but explicit invocation of 'git repack -adf' may improve performance.
CONFIGURATION VARIABLES
-----------------------
The following config settings can be used to modify 'git p4' behavior.
They all are in the 'git-p4' section.
General variables
~~~~~~~~~~~~~~~~~
git-p4.user::
User specified as an option to all p4 commands, with '-u <user>'.
The environment variable 'P4USER' can be used instead.
git-p4.password::
Password specified as an option to all p4 commands, with
'-P <password>'.
The environment variable 'P4PASS' can be used instead.
git-p4.port::
Port specified as an option to all p4 commands, with
'-p <port>'.
The environment variable 'P4PORT' can be used instead.
git-p4.host::
Host specified as an option to all p4 commands, with
'-h <host>'.
The environment variable 'P4HOST' can be used instead.
git-p4.client::
Client specified as an option to all p4 commands, with
'-c <client>', including the client spec.
Clone and sync variables
~~~~~~~~~~~~~~~~~~~~~~~~
git-p4.syncFromOrigin::
Because importing commits from other Git repositories is much faster
than importing them from p4, a mechanism exists to find p4 changes
first in Git remotes. If branches exist under 'refs/remote/origin/p4',
those will be fetched and used when syncing from p4. This
variable can be set to 'false' to disable this behavior.
git-p4.branchUser::
One phase in branch detection involves looking at p4 branches
to find new ones to import. By default, all branches are
inspected. This option limits the search to just those owned
by the single user named in the variable.
git-p4.branchList::
List of branches to be imported when branch detection is
enabled. Each entry should be a pair of branch names separated
by a colon (:). This example declares that both branchA and
branchB were created from main:
+
-------------
git config git-p4.branchList main:branchA
git config --add git-p4.branchList main:branchB
-------------
git-p4.ignoredP4Labels::
List of p4 labels to ignore. This is built automatically as
unimportable labels are discovered.
git-p4.importLabels::
Import p4 labels into git, as per --import-labels.
git-p4.labelImportRegexp::
Only p4 labels matching this regular expression will be imported. The
default value is '[a-zA-Z0-9_\-.]+$'.
git-p4.useClientSpec::
Specify that the p4 client spec should be used to identify p4
depot paths of interest. This is equivalent to specifying the
option '--use-client-spec'. See the "CLIENT SPEC" section above.
This variable is a boolean, not the name of a p4 client.
Submit variables
~~~~~~~~~~~~~~~~
git-p4.detectRenames::
Detect renames. See linkgit:git-diff[1]. This can be true,
false, or a score as expected by 'git diff -M'.
git-p4.detectCopies::
Detect copies. See linkgit:git-diff[1]. This can be true,
false, or a score as expected by 'git diff -C'.
git-p4.detectCopiesHarder::
Detect copies harder. See linkgit:git-diff[1]. A boolean.
git-p4.preserveUser::
On submit, re-author changes to reflect the Git author,
regardless of who invokes 'git p4 submit'.
git-p4.allowMissingP4Users::
When 'preserveUser' is true, 'git p4' normally dies if it
cannot find an author in the p4 user map. This setting
submits the change regardless.
git-p4.skipSubmitEdit::
The submit process invokes the editor before each p4 change
is submitted. If this setting is true, though, the editing
step is skipped.
git-p4.skipSubmitEditCheck::
After editing the p4 change message, 'git p4' makes sure that
the description really was changed by looking at the file
modification time. This option disables that test.
git-p4.allowSubmit::
By default, any branch can be used as the source for a 'git p4
submit' operation. This configuration variable, if set, permits only
the named branches to be used as submit sources. Branch names
must be the short names (no "refs/heads/"), and should be
separated by commas (","), with no spaces.
git-p4.skipUserNameCheck::
If the user running 'git p4 submit' does not exist in the p4
user map, 'git p4' exits. This option can be used to force
submission regardless.
git-p4.attemptRCSCleanup::
If enabled, 'git p4 submit' will attempt to cleanup RCS keywords
($Header$, etc). These would otherwise cause merge conflicts and prevent
the submit going ahead. This option should be considered experimental at
present.
git-p4.exportLabels::
Export Git tags to p4 labels, as per --export-labels.
git-p4.labelExportRegexp::
Only p4 labels matching this regular expression will be exported. The
default value is '[a-zA-Z0-9_\-.]+$'.
git-p4.conflict::
Specify submit behavior when a conflict with p4 is found, as per
--conflict. The default behavior is 'ask'.
IMPLEMENTATION DETAILS
----------------------
* Changesets from p4 are imported using Git fast-import.
* Cloning or syncing does not require a p4 client; file contents are
collected using 'p4 print'.
* Submitting requires a p4 client, which is not in the same location
as the Git repository. Patches are applied, one at a time, to
this p4 client and submitted from there.
* Each commit imported by 'git p4' has a line at the end of the log
message indicating the p4 depot location and change number. This
line is used by later 'git p4 sync' operations to know which p4
changes are new.
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...