1
0
Fork 0
mirror of https://we.phorge.it/source/arcanist.git synced 2024-12-04 12:42:42 +01:00
phorge-arcanist/src/repository/api/ArcanistRepositoryAPI.php

647 lines
17 KiB
PHP
Raw Normal View History

2011-01-10 00:22:25 +01:00
<?php
/**
* Interfaces with the VCS in the working copy.
*
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
* @task status Path Status
* @group workingcopy
*/
2011-01-10 00:22:25 +01:00
abstract class ArcanistRepositoryAPI {
const FLAG_MODIFIED = 1;
const FLAG_ADDED = 2;
const FLAG_DELETED = 4;
const FLAG_UNTRACKED = 8;
const FLAG_CONFLICT = 16;
const FLAG_MISSING = 32;
const FLAG_UNSTAGED = 64;
const FLAG_UNCOMMITTED = 128;
const FLAG_EXTERNALS = 256;
// Occurs in SVN when you replace a file with a directory without telling
// SVN about it.
2011-01-10 00:22:25 +01:00
const FLAG_OBSTRUCTED = 512;
// Occurs in SVN when an update was interrupted or failed, e.g. you ^C'd it.
const FLAG_INCOMPLETE = 1024;
2011-01-10 00:22:25 +01:00
protected $path;
protected $diffLinesOfContext = 0x7FFF;
private $baseCommitExplanation = '???';
private $configurationManager;
Add a DSL for selecting base commits Summary: New optional mode. If you set 'base' in local, project or global config or pass '--base' to 'arc diff' or 'arc which', it switches to DSL mode. In DSL mode, lists of rules from args, local, project and global config are resolved, in that order. Rules can manipulate the rule machine or resolve into actual commits. Provides support for some 'arc' rules (mostly machine manipulation) and 'git' rules (symbolic ref and merge-base). Test Plan: Ran unit tests. Also: ```$ arc which --show-base --base 'arc:prompt' Against which commit? HEAD HEAD $ arc which --show-base --base 'git:HEAD' HEAD $ arc which --show-base --base 'git:fake' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'git:origin/master' origin/master $ arc which --show-base --base 'git:upstream' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'literal:derp' derp $ arc which --show-base --base 'arc:halt' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc set-config --local base git:origin/master Set key 'base' = 'git:origin/master' in local config. $ arc which --show-base origin/master $ arc which --show-base --base 'git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:yield, git:HEAD^' origin/master $ arc which --show-base --base 'arc:global, git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:global, git:merge-base(origin/master)' 3f4f8992fba8d1f142974da36a82bae900e247c0``` Reviewers: dschleimer, vrana Reviewed By: dschleimer CC: aran Maniphest Tasks: T1233 Differential Revision: https://secure.phabricator.com/D2748
2012-06-15 23:01:28 +02:00
private $baseCommitArgumentRules;
2011-01-10 00:22:25 +01:00
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
private $uncommittedStatusCache;
private $commitRangeStatusCache;
private $symbolicBaseCommit;
private $resolvedBaseCommit;
2011-01-10 00:22:25 +01:00
abstract public function getSourceControlSystemName();
public function getDiffLinesOfContext() {
return $this->diffLinesOfContext;
}
public function setDiffLinesOfContext($lines) {
$this->diffLinesOfContext = $lines;
return $this;
}
public function getWorkingCopyIdentity() {
return $this->configurationManager->getWorkingCopyIdentity();
}
public function getConfigurationManager() {
return $this->configurationManager;
}
public static function newAPIFromConfigurationManager(
ArcanistConfigurationManager $configuration_manager) {
$working_copy = $configuration_manager->getWorkingCopyIdentity();
if (!$working_copy) {
throw new Exception(
pht(
"Trying to create a RepositoryAPI without a working copy!"));
}
2011-01-10 00:22:25 +01:00
$root = $working_copy->getProjectRoot();
switch ($working_copy->getVCSType()) {
case 'svn':
$api = new ArcanistSubversionAPI($root);
break;
case 'hg':
$api = new ArcanistMercurialAPI($root);
break;
case 'git':
$api = new ArcanistGitAPI($root);
break;
default:
throw new Exception(
pht(
"The current working directory is not part of a working copy for ".
"a supported version control system (Git, Subversion or ".
"Mercurial)."));
}
$api->configurationManager = $configuration_manager;
return $api;
2011-01-10 00:22:25 +01:00
}
public function __construct($path) {
2011-01-10 00:22:25 +01:00
$this->path = $path;
}
public function getPath($to_file = null) {
if ($to_file !== null) {
return $this->path.DIRECTORY_SEPARATOR.
ltrim($to_file, DIRECTORY_SEPARATOR);
2011-01-10 00:22:25 +01:00
} else {
return $this->path.DIRECTORY_SEPARATOR;
2011-01-10 00:22:25 +01:00
}
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/* -( Path Status )-------------------------------------------------------- */
abstract protected function buildUncommittedStatus();
abstract protected function buildCommitRangeStatus();
/**
* Get a list of uncommitted paths in the working copy that have been changed
* or are affected by other status effects, like conflicts or untracked
* files.
*
* Convenience methods @{method:getUntrackedChanges},
* @{method:getUnstagedChanges}, @{method:getUncommittedChanges},
* @{method:getMergeConflicts}, and @{method:getIncompleteChanges} allow
* simpler selection of paths in a specific state.
*
* This method returns a map of paths to bitmasks with status, using
* `FLAG_` constants. For example:
*
* array(
* 'some/uncommitted/file.txt' => ArcanistRepositoryAPI::FLAG_UNSTAGED,
* );
*
* A file may be in several states. Not all states are possible with all
* version control systems.
*
* @return map<string, bitmask> Map of paths, see above.
* @task status
*/
final public function getUncommittedStatus() {
if ($this->uncommittedStatusCache === null) {
2013-01-29 00:15:18 +01:00
$status = $this->buildUncommittedStatus();
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
ksort($status);
$this->uncommittedStatusCache = $status;
}
return $this->uncommittedStatusCache;
2011-01-10 00:22:25 +01:00
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/**
* @task status
*/
final public function getUntrackedChanges() {
return $this->getUncommittedPathsWithMask(self::FLAG_UNTRACKED);
}
/**
* @task status
*/
final public function getUnstagedChanges() {
return $this->getUncommittedPathsWithMask(self::FLAG_UNSTAGED);
2011-01-10 00:22:25 +01:00
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/**
* @task status
*/
final public function getUncommittedChanges() {
return $this->getUncommittedPathsWithMask(self::FLAG_UNCOMMITTED);
2011-01-10 00:22:25 +01:00
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/**
* @task status
*/
final public function getMergeConflicts() {
return $this->getUncommittedPathsWithMask(self::FLAG_CONFLICT);
2011-01-10 00:22:25 +01:00
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/**
* @task status
*/
final public function getIncompleteChanges() {
return $this->getUncommittedPathsWithMask(self::FLAG_INCOMPLETE);
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/**
* @task status
*/
final public function getMissingChanges() {
return $this->getUncommittedPathsWithMask(self::FLAG_MISSING);
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/**
* @task status
*/
private function getUncommittedPathsWithMask($mask) {
2011-01-10 00:22:25 +01:00
$match = array();
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
foreach ($this->getUncommittedStatus() as $path => $flags) {
2011-01-10 00:22:25 +01:00
if ($flags & $mask) {
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
$match[] = $path;
2011-01-10 00:22:25 +01:00
}
}
return $match;
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/**
* Get a list of paths affected by the commits in the current commit range.
*
* See @{method:getUncommittedStatus} for a description of the return value.
*
* @return map<string, bitmask> Map from paths to status.
* @task status
*/
final public function getCommitRangeStatus() {
if ($this->commitRangeStatusCache === null) {
$status = $this->buildCommitRangeStatus();
ksort($status);
$this->commitRangeStatusCache = $status;
}
return $this->commitRangeStatusCache;
}
/**
* Get a list of paths affected by commits in the current commit range, or
* uncommitted changes in the working copy. See @{method:getUncommittedStatus}
* or @{method:getCommitRangeStatus} to retreive smaller parts of the status.
*
* See @{method:getUncommittedStatus} for a description of the return value.
*
* @return map<string, bitmask> Map from paths to status.
* @task status
*/
final public function getWorkingCopyStatus() {
$range_status = $this->getCommitRangeStatus();
$uncommitted_status = $this->getUncommittedStatus();
$result = new PhutilArrayWithDefaultValue($range_status);
foreach ($uncommitted_status as $path => $mask) {
$result[$path] |= $mask;
}
$result = $result->toArray();
ksort($result);
return $result;
}
/**
* Drops caches after changes to the working copy. By default, some queries
* against the working copy are cached. They
*
* @return this
* @task status
*/
final public function reloadWorkingCopy() {
$this->uncommittedStatusCache = null;
$this->commitRangeStatusCache = null;
$this->didReloadWorkingCopy();
$this->reloadCommitRange();
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
return $this;
}
/**
* Hook for implementations to dirty working copy caches after the working
* copy has been updated.
*
* @return this
* @task status
*/
protected function didReloadWorkingCopy() {
return;
}
/**
* Fetches the original file data for each path provided.
*
* @return map<string, string> Map from path to file data.
*/
public function getBulkOriginalFileData($paths) {
$filedata = array();
foreach ($paths as $path) {
$filedata[$path] = $this->getOriginalFileData($path);
}
return $filedata;
}
/**
* Fetches the current file data for each path provided.
*
* @return map<string, string> Map from path to file data.
*/
public function getBulkCurrentFileData($paths) {
$filedata = array();
foreach ($paths as $path) {
$filedata[$path] = $this->getCurrentFileData($path);
}
return $filedata;
}
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
/**
* @return Traversable
*/
abstract public function getAllFiles();
2011-01-10 00:22:25 +01:00
abstract public function getBlame($path);
Refactor getWorkingCopyStatus() into getUncommittedStatus() and getCommitRangeStatus() Summary: See discussion in D4049. The getWorkingCopyStatus() method gets called from requireCleanWorkingCopy() in a lot of places, which triggers resolution of the base of the commit range. This is unnecessary; we do not need to examine the base commit in order to determine whether the working copy is dirty or not. This causes problems, in D4049 and elsewhere (we currently have a lot of fluff calls to setDefaultBaseCommit() in workflows which need to call requireCleanWorkingCopy() but do not ever use commit ranges, such as `arc patch`). This is mostly an artifact of SVN, where the "commit range" and "uncommitted stuff in the working copy" are always the same. - Split the method into two status methods: getUncommittedStatus() (uncommitted stuff in the working copy, required by requireCleanWorkingCopy()) and getCommitRangeStatus() (committed stuff in the commit range). - Lift caching out of the implementations into the base class. - Dirty the cache after we commit. This doesn't do anything useful on its own and creates one caching problem (`commitRangeStatusCache` is not invalidated when the commit range changes because of `setBaseCommit()` or similar) but I wanted to break things apart here. I won't land it until there's a more complete picture. This creates a minor performance regression in git and hg (we run less stuff in parallel than previously) but all the commands should be disk-bound anyway and the regression should be minor. It prevents a larger regression in `hg` in D4049, and lets us do less work to arrive at common error states (dirty working copy). We can examine perf at the end of this change sequence. Test Plan: Ran unit tests, various `arc` commands. Reviewers: vrana Reviewed By: vrana CC: aran Differential Revision: https://secure.phabricator.com/D4095
2012-12-17 21:53:28 +01:00
2011-01-10 00:22:25 +01:00
abstract public function getRawDiffText($path);
abstract public function getOriginalFileData($path);
abstract public function getCurrentFileData($path);
abstract public function getLocalCommitInformation();
abstract public function getSourceControlBaseRevision();
abstract public function getCanonicalRevisionName($string);
abstract public function getBranchName();
abstract public function getSourceControlPath();
abstract public function isHistoryDefaultImmutable();
abstract public function supportsAmend();
abstract public function getWorkingCopyRevision();
abstract public function updateWorkingCopy();
abstract public function getMetadataPath();
abstract public function loadWorkingCopyDifferentialRevisions(
ConduitClient $conduit,
array $query);
public function getUnderlyingWorkingCopyRevision() {
return $this->getWorkingCopyRevision();
}
public function getChangedFiles($since_commit) {
throw new ArcanistCapabilityNotSupportedException($this);
}
public function getAuthor() {
throw new ArcanistCapabilityNotSupportedException($this);
}
public function addToCommit(array $paths) {
throw new ArcanistCapabilityNotSupportedException($this);
}
abstract public function supportsLocalCommits();
public function doCommit($message) {
throw new ArcanistCapabilityNotSupportedException($this);
}
public function amendCommit($message = null) {
throw new ArcanistCapabilityNotSupportedException($this);
}
public function getAllBranches() {
// TODO: Implement for Mercurial/SVN and make abstract.
return array();
}
public function hasLocalCommit($commit) {
throw new ArcanistCapabilityNotSupportedException($this);
}
2012-07-01 20:06:05 +02:00
public function getCommitMessage($commit) {
throw new ArcanistCapabilityNotSupportedException($this);
}
public function getCommitSummary($commit) {
throw new ArcanistCapabilityNotSupportedException($this);
}
public function getAllLocalChanges() {
throw new ArcanistCapabilityNotSupportedException($this);
}
abstract public function supportsLocalBranchMerge();
public function performLocalBranchMerge($branch, $message) {
throw new ArcanistCapabilityNotSupportedException($this);
}
public function getFinalizedRevisionMessage() {
throw new ArcanistCapabilityNotSupportedException($this);
}
2012-07-01 20:06:05 +02:00
public function execxLocal($pattern /* , ... */) {
$args = func_get_args();
return $this->buildLocalFuture($args)->resolvex();
}
2012-07-01 20:06:05 +02:00
public function execManualLocal($pattern /* , ... */) {
$args = func_get_args();
return $this->buildLocalFuture($args)->resolve();
}
2012-07-01 20:06:05 +02:00
public function execFutureLocal($pattern /* , ... */) {
$args = func_get_args();
return $this->buildLocalFuture($args);
}
abstract protected function buildLocalFuture(array $argv);
public function canStashChanges() {
return false;
}
public function stashChanges() {
throw new ArcanistCapabilityNotSupportedException($this);
}
public function unstashChanges() {
throw new ArcanistCapabilityNotSupportedException($this);
}
/* -( Scratch Files )------------------------------------------------------ */
/**
* Try to read a scratch file, if it exists and is readable.
*
* @param string Scratch file name.
* @return mixed String for file contents, or false for failure.
* @task scratch
*/
public function readScratchFile($path) {
$full_path = $this->getScratchFilePath($path);
if (!$full_path) {
return false;
}
if (!Filesystem::pathExists($full_path)) {
return false;
}
try {
$result = Filesystem::readFile($full_path);
} catch (FilesystemException $ex) {
return false;
}
return $result;
}
/**
* Try to write a scratch file, if there's somewhere to put it and we can
* write there.
*
* @param string Scratch file name to write.
* @param string Data to write.
* @return bool True on success, false on failure.
* @task scratch
*/
public function writeScratchFile($path, $data) {
$dir = $this->getScratchFilePath('');
if (!$dir) {
return false;
}
if (!Filesystem::pathExists($dir)) {
try {
Filesystem::createDirectory($dir);
} catch (Exception $ex) {
return false;
}
}
try {
Filesystem::writeFile($this->getScratchFilePath($path), $data);
} catch (FilesystemException $ex) {
return false;
}
return true;
}
/**
* Try to remove a scratch file.
*
* @param string Scratch file name to remove.
* @return bool True if the file was removed successfully.
* @task scratch
*/
public function removeScratchFile($path) {
$full_path = $this->getScratchFilePath($path);
if (!$full_path) {
return false;
}
try {
Filesystem::remove($full_path);
} catch (FilesystemException $ex) {
return false;
}
return true;
}
/**
* Get a human-readable description of the scratch file location.
*
* @param string Scratch file name.
* @return mixed String, or false on failure.
* @task scratch
*/
public function getReadableScratchFilePath($path) {
$full_path = $this->getScratchFilePath($path);
if ($full_path) {
return Filesystem::readablePath(
$full_path,
$this->getPath());
} else {
return false;
}
}
/**
* Get the path to a scratch file, if possible.
*
* @param string Scratch file name.
* @return mixed File path, or false on failure.
* @task scratch
*/
public function getScratchFilePath($path) {
$new_scratch_path = Filesystem::resolvePath(
'arc',
$this->getMetadataPath());
static $checked = false;
if (!$checked) {
$checked = true;
$old_scratch_path = $this->getPath('.arc');
// we only want to do the migration once
// unfortunately, people have checked in .arc directories which
// means that the old one may get recreated after we delete it
if (Filesystem::pathExists($old_scratch_path) &&
!Filesystem::pathExists($new_scratch_path)) {
Filesystem::createDirectory($new_scratch_path);
$existing_files = Filesystem::listDirectory($old_scratch_path, true);
foreach ($existing_files as $file) {
$new_path = Filesystem::resolvePath($file, $new_scratch_path);
$old_path = Filesystem::resolvePath($file, $old_scratch_path);
Filesystem::writeFile(
$new_path,
Filesystem::readFile($old_path));
}
Filesystem::remove($old_scratch_path);
}
}
return Filesystem::resolvePath($path, $new_scratch_path);
}
Add a DSL for selecting base commits Summary: New optional mode. If you set 'base' in local, project or global config or pass '--base' to 'arc diff' or 'arc which', it switches to DSL mode. In DSL mode, lists of rules from args, local, project and global config are resolved, in that order. Rules can manipulate the rule machine or resolve into actual commits. Provides support for some 'arc' rules (mostly machine manipulation) and 'git' rules (symbolic ref and merge-base). Test Plan: Ran unit tests. Also: ```$ arc which --show-base --base 'arc:prompt' Against which commit? HEAD HEAD $ arc which --show-base --base 'git:HEAD' HEAD $ arc which --show-base --base 'git:fake' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'git:origin/master' origin/master $ arc which --show-base --base 'git:upstream' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'literal:derp' derp $ arc which --show-base --base 'arc:halt' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc set-config --local base git:origin/master Set key 'base' = 'git:origin/master' in local config. $ arc which --show-base origin/master $ arc which --show-base --base 'git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:yield, git:HEAD^' origin/master $ arc which --show-base --base 'arc:global, git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:global, git:merge-base(origin/master)' 3f4f8992fba8d1f142974da36a82bae900e247c0``` Reviewers: dschleimer, vrana Reviewed By: dschleimer CC: aran Maniphest Tasks: T1233 Differential Revision: https://secure.phabricator.com/D2748
2012-06-15 23:01:28 +02:00
/* -( Base Commits )------------------------------------------------------- */
abstract public function supportsCommitRanges();
final public function setBaseCommit($symbolic_commit) {
if (!$this->supportsCommitRanges()) {
throw new ArcanistCapabilityNotSupportedException($this);
}
$this->symbolicBaseCommit = $symbolic_commit;
$this->reloadCommitRange();
return $this;
}
final public function getBaseCommit() {
if (!$this->supportsCommitRanges()) {
throw new ArcanistCapabilityNotSupportedException($this);
}
if ($this->resolvedBaseCommit === null) {
$commit = $this->buildBaseCommit($this->symbolicBaseCommit);
$this->resolvedBaseCommit = $commit;
}
return $this->resolvedBaseCommit;
}
final public function reloadCommitRange() {
$this->resolvedBaseCommit = null;
$this->baseCommitExplanation = null;
$this->didReloadCommitRange();
return $this;
}
protected function didReloadCommitRange() {
return;
}
protected function buildBaseCommit($symbolic_commit) {
throw new ArcanistCapabilityNotSupportedException($this);
}
Add a DSL for selecting base commits Summary: New optional mode. If you set 'base' in local, project or global config or pass '--base' to 'arc diff' or 'arc which', it switches to DSL mode. In DSL mode, lists of rules from args, local, project and global config are resolved, in that order. Rules can manipulate the rule machine or resolve into actual commits. Provides support for some 'arc' rules (mostly machine manipulation) and 'git' rules (symbolic ref and merge-base). Test Plan: Ran unit tests. Also: ```$ arc which --show-base --base 'arc:prompt' Against which commit? HEAD HEAD $ arc which --show-base --base 'git:HEAD' HEAD $ arc which --show-base --base 'git:fake' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'git:origin/master' origin/master $ arc which --show-base --base 'git:upstream' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'literal:derp' derp $ arc which --show-base --base 'arc:halt' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc set-config --local base git:origin/master Set key 'base' = 'git:origin/master' in local config. $ arc which --show-base origin/master $ arc which --show-base --base 'git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:yield, git:HEAD^' origin/master $ arc which --show-base --base 'arc:global, git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:global, git:merge-base(origin/master)' 3f4f8992fba8d1f142974da36a82bae900e247c0``` Reviewers: dschleimer, vrana Reviewed By: dschleimer CC: aran Maniphest Tasks: T1233 Differential Revision: https://secure.phabricator.com/D2748
2012-06-15 23:01:28 +02:00
public function getBaseCommitExplanation() {
return $this->baseCommitExplanation;
}
public function setBaseCommitExplanation($explanation) {
$this->baseCommitExplanation = $explanation;
return $this;
}
public function resolveBaseCommitRule($rule, $source) {
return null;
}
public function setBaseCommitArgumentRules($base_commit_argument_rules) {
$this->baseCommitArgumentRules = $base_commit_argument_rules;
return $this;
}
public function getBaseCommitArgumentRules() {
return $this->baseCommitArgumentRules;
}
public function resolveBaseCommit() {
$base_commit_rules = array(
'runtime' => $this->getBaseCommitArgumentRules(),
'local' => '',
'project' => '',
'user' => '',
'system' => '',
);
$all_sources = $this->configurationManager->getConfigFromAllSources('base');
$base_commit_rules = $all_sources + $base_commit_rules;
Add a DSL for selecting base commits Summary: New optional mode. If you set 'base' in local, project or global config or pass '--base' to 'arc diff' or 'arc which', it switches to DSL mode. In DSL mode, lists of rules from args, local, project and global config are resolved, in that order. Rules can manipulate the rule machine or resolve into actual commits. Provides support for some 'arc' rules (mostly machine manipulation) and 'git' rules (symbolic ref and merge-base). Test Plan: Ran unit tests. Also: ```$ arc which --show-base --base 'arc:prompt' Against which commit? HEAD HEAD $ arc which --show-base --base 'git:HEAD' HEAD $ arc which --show-base --base 'git:fake' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'git:origin/master' origin/master $ arc which --show-base --base 'git:upstream' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'literal:derp' derp $ arc which --show-base --base 'arc:halt' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc set-config --local base git:origin/master Set key 'base' = 'git:origin/master' in local config. $ arc which --show-base origin/master $ arc which --show-base --base 'git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:yield, git:HEAD^' origin/master $ arc which --show-base --base 'arc:global, git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:global, git:merge-base(origin/master)' 3f4f8992fba8d1f142974da36a82bae900e247c0``` Reviewers: dschleimer, vrana Reviewed By: dschleimer CC: aran Maniphest Tasks: T1233 Differential Revision: https://secure.phabricator.com/D2748
2012-06-15 23:01:28 +02:00
$parser = new ArcanistBaseCommitParser($this);
$commit = $parser->resolveBaseCommit($base_commit_rules);
Add a DSL for selecting base commits Summary: New optional mode. If you set 'base' in local, project or global config or pass '--base' to 'arc diff' or 'arc which', it switches to DSL mode. In DSL mode, lists of rules from args, local, project and global config are resolved, in that order. Rules can manipulate the rule machine or resolve into actual commits. Provides support for some 'arc' rules (mostly machine manipulation) and 'git' rules (symbolic ref and merge-base). Test Plan: Ran unit tests. Also: ```$ arc which --show-base --base 'arc:prompt' Against which commit? HEAD HEAD $ arc which --show-base --base 'git:HEAD' HEAD $ arc which --show-base --base 'git:fake' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'git:origin/master' origin/master $ arc which --show-base --base 'git:upstream' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc which --show-base --base 'literal:derp' derp $ arc which --show-base --base 'arc:halt' Usage Exception: None of the rules in your 'base' configuration matched a valid commit. Adjust rules or specify which commit you want to use explicitly. $ arc set-config --local base git:origin/master Set key 'base' = 'git:origin/master' in local config. $ arc which --show-base origin/master $ arc which --show-base --base 'git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:yield, git:HEAD^' origin/master $ arc which --show-base --base 'arc:global, git:HEAD^' HEAD^ $ arc which --show-base --base 'arc:global, git:merge-base(origin/master)' 3f4f8992fba8d1f142974da36a82bae900e247c0``` Reviewers: dschleimer, vrana Reviewed By: dschleimer CC: aran Maniphest Tasks: T1233 Differential Revision: https://secure.phabricator.com/D2748
2012-06-15 23:01:28 +02:00
return $commit;
}
2011-01-10 00:22:25 +01:00
}