Differential D20467 Diff 48843 src/applications/repository/storage/PhabricatorRepositoryCommit.php

Changeset View

Standalone View

src/applications/repository/storage/PhabricatorRepositoryCommit.php

Show All 27 Lines	final class PhabricatorRepositoryCommit
protected $epoch;		protected $epoch;
protected $authorPHID;		protected $authorPHID;
protected $auditStatus = DiffusionCommitAuditStatus::NONE;		protected $auditStatus = DiffusionCommitAuditStatus::NONE;
protected $summary = '';		protected $summary = '';
protected $importStatus = 0;		protected $importStatus = 0;

const IMPORTED_MESSAGE = 1;		const IMPORTED_MESSAGE = 1;
const IMPORTED_CHANGE = 2;		const IMPORTED_CHANGE = 2;
const IMPORTED_OWNERS = 4;		const IMPORTED_PUBLISH = 8;
const IMPORTED_HERALD = 8;		const IMPORTED_ALL = 11;
		amckinleyUnsubmitted Not Done Inline Actions What happens to commits currently sitting around with `importStatus IN (IMPORTED_OWNERS, IMPORTED_HERALD)` when this update gets applied? I guess we'll just end up running the publish step again, which is fine? amckinley: What happens to commits currently sitting around with `importStatus IN (IMPORTED_OWNERS…
		epriestleyAuthorUnsubmitted Done Inline Actions If the next queued import step for a commit is: "Message" or "Change": Things will work fine, they'll just move from "Change" to "Publish". "Owners" or "Herald": The task classes no longer exist, so these will tasks will permanently. The commits will never fully import. Administrators will need to `bin/repository reparse` if they want these steps to finish. I'll call this out in the changelog ("Finish importing any outstanding repositories before upgrading"), but even if users ignore this advice the number of impacted commits should be small, and skipping some parse steps for a handful of commits normally isn't a problem. epriestley: If the next queued import step for a commit is: - "Message" or "Change": Things will work…
		amckinleyUnsubmitted Not Done Inline Actions Don't we need a migration for this? Won't the DB suddenly have a bunch of old `PhabricatorRepositoryCommit` with `importStatus = 15` and a bunch of new ones with `importStatus = 11`? amckinley: Don't we need a migration for this? Won't the DB suddenly have a bunch of old…
		epriestleyAuthorUnsubmitted Done Inline Actions Older commits will have an extra bit (4) in the bitfield, but that doesn't hurt anything. All tests against the bitfield examine specific bits and this field is purely internal. For the sake of cleanliness, we could wipe this bit from memory by setting the field to `field & ~4`, but that seems fairly low-value and would make reverting this change more difficult/dangerous. epriestley: Older commits will have an extra bit (4) in the bitfield, but that doesn't hurt anything. All…
const IMPORTED_ALL = 15;

const IMPORTED_CLOSEABLE = 1024;		const IMPORTED_CLOSEABLE = 1024;
const IMPORTED_UNREACHABLE = 2048;		const IMPORTED_UNREACHABLE = 2048;

private $commitData = self::ATTACHABLE;		private $commitData = self::ATTACHABLE;
private $audits = self::ATTACHABLE;		private $audits = self::ATTACHABLE;
private $repository = self::ATTACHABLE;		private $repository = self::ATTACHABLE;
private $customFields = self::ATTACHABLE;		private $customFields = self::ATTACHABLE;
▲ Show 20 Lines • Show All 855 Lines • Show Last 20 Lines