Summary: Ref T603. Makes the majority of reads policy aware (and pretty much all the important ones).
Test Plan:
- Created a comment with `differential.createcomment`.
- Created a new revision with `arc diff` in order to exercise `differential.creatediff`.
- Created an inline comment with `differential.createinline`.
- Added a comment to a revision.
- Edited an inline comment.
- Edited a revision.
- Wrote "Depends on ..." in a summary, saved, verified link was created.
- Browsed a file in Diffusion.
- Got past the code I changed in the Releeph request thing.
- Edited a Releeph request.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T603
Differential Revision: https://secure.phabricator.com/D7136
Summary: Ref T603. This swaps almost all queries against the repository table over to be policy aware.
Test Plan:
- Made an audit comment on a commit.
- Ran `save_lint.php`.
- Looked up a commit with `diffusion.getcommits`.
- Looked up lint messages with `diffusion.getlintmessages`.
- Clicked an external/submodule in Diffusion.
- Viewed main lint and repository lint in Diffusion.
- Completed and validated Owners paths in Owners.
- Executed dry runs via Herald.
- Queried for package owners with `owners.query`.
- Viewed Owners package.
- Edited Owners package.
- Viewed Owners package list.
- Executed `repository.query`.
- Viewed "Repository" tool repository list.
- Edited Arcanist project.
- Hit "Delete" on repository (this just tells you to use the CLI).
- Created a repository.
- Edited a repository.
- Ran `bin/repository list`.
- Ran `bin/search index rGTESTff45d13dffcfb3ea85b03aac8cc36251cacdf01c`
- Pushed and parsed a commit.
- Skipped all the Drydock stuff, as it it's hard to test and isn't normally reachable.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T603
Differential Revision: https://secure.phabricator.com/D7132
Summary:
Fixes T3821. Maybe. The existing code seemed to have a bug and actually return the //commit phid//. Judging by the function name this is not intended.
Also, sorry to step on toes here -- I thought no one was assigned and was curious about loadRelativeEdges and here we are...
Test Plan: lots of logic here as I have no idea how to use Releeph.
Reviewers: epriestley
Reviewed By: epriestley
CC: Korvin, aran
Maniphest Tasks: T3821
Differential Revision: https://secure.phabricator.com/D6967
Summary: Ref T603. Killing this class is cool because the classes that replace it are policy-aware. Tried to keep my wits about me as I did this and fixed a few random things along the way. (Ones I remember right now are pulling a query outside of a foreach loop in Releeph and fixing the text in UIExample to note that the ace of hearts if "a powerful" card and not the "most powerful" card (Q of spades gets that honor IMO))
Test Plan: tested the first few changes (execute, executeOne X handle, object) then got real mechanical / careful with the other changes.
Reviewers: epriestley
Reviewed By: epriestley
CC: Korvin, aran, FacebookPOC
Maniphest Tasks: T603
Differential Revision: https://secure.phabricator.com/D6941
Summary:
Ref T3599
Go through everything, grep a bit, replace some bits.
Test Plan: Navigate around a bit
Reviewers: epriestley
Reviewed By: epriestley
CC: Korvin, aran
Maniphest Tasks: T3599
Differential Revision: https://secure.phabricator.com/D6871
Summary:
We can get this out of PHIDType reasonably in all cases and simplify implementation here.
None of these translate correctly anyway so they're basically debugging/development strings.
Test Plan: `grep`, browsed some transactions
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Differential Revision: https://secure.phabricator.com/D6786
Summary:
Ref T3663. This is a proto-transaction record which is obsoleted by real transactions. It has no UI, so I'm not bothering to retain/migrate the data since there's no regression.
Just get rid of it and all its writers. I'm keeping the table for now in case something crazy uses this somehow, so no data is actually destroyed.
Test Plan: `grep`
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T3663
Differential Revision: https://secure.phabricator.com/D6784
Summary: This has two use sites and no special logic.
Test Plan: `grep`
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Differential Revision: https://secure.phabricator.com/D6783
Summary: Ref T3663. Does what it says on the tin.
Test Plan: Ran `storage upgrade`.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T3663
Differential Revision: https://secure.phabricator.com/D6778
Summary: Ref T3663. This is obsolete code which is used only in this migration, which Facebook has already performed and which isn't relevant for any other installs.
Test Plan: `grep`
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T3663
Differential Revision: https://secure.phabricator.com/D6777
Summary:
Ref T3092.
Releeph's objects basically go like this:
- At the top level, we have Projects (like "www" or "libphutil")
- Each project has Branches (like "LATEST" or "v1.1.3")
- Each branch has Requests (like pull requests, e.g. "please merge commit X into branch Y (in project Z)")
Currently, there's no real "project detail" or "branch detail" page. Instead, we have a search results page for their contained objects. That is, the "project detail" page shows a list of branches in the project, using ApplicationSearch.
This means that operations like "edit" and "deactivate" are one level up, on the respective list pages.
Instead, move details onto the detail pages. This gives us more room for actions and information, and simplifies the list views.
Basically, these are "detail pages" where the object content is a search interface. We do something simliar to this in Phame right now, although it's messier there (no ApplicationSearch yet).
@chad, you might have some ideas here. Roughly, the design question is "How should we present an object's detail view when its content is really a search interface (Phame Blog for Posts, Releeph Project for Branches)?"
I think the simple approach I've taken here (see screenshot) gives us reasonable results, but overall it's something we haven't done much or done too much thinking about, I think.
Test Plan: {F54774}
Reviewers: btrahan
Reviewed By: btrahan
CC: chad, aran
Maniphest Tasks: T3092
Differential Revision: https://secure.phabricator.com/D6771
Summary:
Fixes T3661. Ref T3718. This makes Releeph custom fields extend PhabricatorCustomField so we can start moving over other pieces of infrastructure (rendering, storage, etc) to run through the same pathways. It's roughly the minimum amount of work required to be able to move forward.
NOTE: This removes per-project custom field selectors. Fields are now configured for an entire install. My understanding is that Facebook does not use this feature, and modern field infrastructure has moved away from selectors.
Test Plan: Viewed and edited projects, branches, and requests in Releeph. Grepped for removed config. Grepped for `field_selector`.
Reviewers: btrahan
Reviewed By: btrahan
CC: LegNeato, aran
Maniphest Tasks: T3661, T3718
Differential Revision: https://secure.phabricator.com/D6750
Summary: Ref T3656. Releeph denormalizes branch cut point identifiers into Branch objects, but this information isn't useful or used for sorting, filtering, or enforcing unique constraints. Instead, derive it via noramlized pathways from the `cutPointCommitPHID`.
Test Plan: Ran storage upgrade. Ran `releephwork.getbranch` and `releeph.getbranches`. Grepped for `cutPointCommitIdentifier`.
Reviewers: btrahan
Reviewed By: btrahan
CC: LegNeato, aran
Maniphest Tasks: T3656
Differential Revision: https://secure.phabricator.com/D6636
Summary:
Fixes T3660. Releeph Projects currently have an unused one-to-one mapping to Phabricator projects. This isn't consistent with other applications and has no integrations or uses. Get rid of it.
NOTE: Waiting for signoff from @legneato on T3660 before pulling the trigger here.
Test Plan: Created and edited Releeph projects. Grepped for references to project ID; there are a dozen or so but they're all either Releeph projects or Arcanist projects.
Reviewers: btrahan
Reviewed By: btrahan
CC: LegNeato, aran
Maniphest Tasks: T3660
Differential Revision: https://secure.phabricator.com/D6635
Summary: Ref T3655. Depends on D6633. This removes the writes and the column.
Test Plan: Created a project, edited a project. Verified the table doesn't have any keys including this column.
Reviewers: btrahan
Reviewed By: btrahan
CC: LegNeato, aran
Maniphest Tasks: T3655
Differential Revision: https://secure.phabricator.com/D6634
Summary:
Ref T3655. ReleephProject currently has both `repositoryID` and `repositoryPHID`, which point to the same object and are reudundant. Get rid of all reads of `repositoryID`.
NOTE: This makes project loads depend on repository loads. The eventual rule here will be that you must be able to see a repository in order to see projects for that repository, which seems like a reasonable rule. We might need to tailor it more than this (e.g., if there are branch read permissions down the line) but this seems like a reasonable minimum.
Test Plan: Grepped for `repositoryID` in `releeph/`. Called `releeph.getbranches`.
Reviewers: btrahan
Reviewed By: btrahan
CC: LegNeato, aran
Maniphest Tasks: T3655
Differential Revision: https://secure.phabricator.com/D6633
Summary:
Ref T2715. Ref T3551. Ref T603. This does a few things, but they're all sort of small:
- We commonly use a `getX()` / `attachX()` pattern, but have very similar code in the `getX()` method every time. Provide a convenience method to make this pattern easier to write.
- We use `willFilterPage()` in many queries, but it currently is called with zero or more results. This means we have a lot of "if no results, return nothing" boilerplate. Make it call only for one or more results.
- Implement `PhabricatorPolicyInterface` on `ReleephBranch`. A branch has the same policy as its project.
- Implement `ReleephBranchQuery`.
- Move the branch PHID type to application PHID infrastructure.
Test Plan: Browsed Releeph. Used `phid.query` to query branch PHIDs.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T603, T2715, T3551
Differential Revision: https://secure.phabricator.com/D6512
Summary: I used the `PhabricatorEnv::getProductionURI()` in too many places to build Releeph URIs. The only places that should need full URIs are the links generated for Releeph emails, and in Conduit responses that link to Releeph objects.
Test Plan:
- Grep for `getProductionURI()` in Releeph, and make sure only sensible, non-DOM building places use it.
- Inspect the Releeph DOM to make sure hrefs etc. are relative.
Reviewers: wez, epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D5999
Summary: This is just a bit of gardening in order to make the responsive-UI diff easier; I'll be putting `getColorFor($status)` type things in this class, following the pattern in `ManiphestTaskStatus`.
Test Plan: Poke around Releeph.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D5893
Summary:
Migrate to `PhabricatorApplicationTransactions` (`ReleephRequestTransactions` applied by `ReleephRequestTransactionalEditor`, instead of `ReleephRequestEvents` created by `ReleephRequestEditor`) and migrate all the old events into transactions. Email is supported in the standard way (no more `ReleephRequestMail`) as well.
This also collapses the Releeph request create and edit controllers into one class, as well as breaking everyone's subject-based mail rules by standardising them (but which should be more easily filtered by looking at headers.)
Test Plan:
* Make requests, then pick them.
* Pick and revert the same request so that discovery happens way after `arc` has told Releeph about what's been happening.
* Try to pick something that fails to pick in a project with pick instructions (and see the instructions are in the email.)
* Load all of FB's Releeph data into my DB and run the `storage upgrade` script.
* Request a commit via the "action" in a Differential revision.
Reviewers: epriestley
Reviewed By: epriestley
CC: epriestley, aran, Korvin, wez
Maniphest Tasks: T3092, T2720
Differential Revision: https://secure.phabricator.com/D5868
Summary: Instead of being able to ask if someone was a pusher or not, ask if they are "authoritative" enough to make decisions about Releeph requests. A person is authoritative if a project has pushers, and they are a pusher, or in the case of pusher-less projects, everyone is authoritative.
Test Plan: Make a request in a project with no pushers (it is immediately ready to be picked) and a project with pushers (where it requires approval.)
Reviewers: wez, epriestley
Reviewed By: epriestley
CC: epriestley, aran
Differential Revision: https://secure.phabricator.com/D5877
Summary: Adding mail-keys; required for `PhabricatorApplicationTransaction` support.
Test Plan: Upgrade an old database with this patch, observe the matrix: {F42620}
Reviewers: wez, epriestley
Reviewed By: epriestley
CC: epriestley, aran
Maniphest Tasks: T2720
Differential Revision: https://secure.phabricator.com/D5852
Summary:
Removing a bunch of cache-style columns from `ReleephRequest`, where it's actually much easier to just load the information at runtime.
This makes sense for migrating to `PhabricatorApplicationTransactions`, where each xaction changes one aspect of a `ReleephRequest` at a time (rather than multiple columns at once.)
Test Plan: Request something, run `arc releeph` and watch the picks, pass on some RQs, run `arc releeph` and watch the reverts.
Reviewers: wez, epriestley
Reviewed By: epriestley
CC: epriestley, aran
Maniphest Tasks: T2720
Differential Revision: https://secure.phabricator.com/D5851
Summary: Not every commit requested in Releeph has to have come from a Differential revision. If you have a Releeph request for a commit without a Differential revision, you currently get this lovely error in the logs: {P824}. This diff silences the howling.
Test Plan: Load a Releeph branch page; observe log silence.
Reviewers: wez, epriestley
Reviewed By: epriestley
CC: epriestley, aran, SEJeff, Korvin
Differential Revision: https://secure.phabricator.com/D5823
Summary:
`ReleephRequest`s contain the PHID for a `PhabricatorRepositoryCommit`, and commits have an edge to a `DifferentialRevision`.
Commits are loaded with the `loadOneRelative()` method that loads the commits for every `ReleephRequest` in a `LiskDAOSet`, but the edges are loaded indivdually. A page with N RQs on it makes one DB query for the commits, but N queries for the `TYPE_COMMIT_HAS_DREV` edges.
This diff uses `loadRelativeEdges` instead to load the edges all in one query.
Test Plan: {F42290}
Reviewers: wez, epriestley
Reviewed By: epriestley
CC: epriestley, vrana, aran
Maniphest Tasks: T2714
Differential Revision: https://secure.phabricator.com/D5820
Summary: Migration doesn't delete differential.revisionPHID but maybe it should?
Test Plan: Reparsed commit, ran the migration, deleted differential.revisionPHID, looked at task with attached commit with attached revision.
Reviewers: epriestley, edward
Reviewed By: epriestley
CC: aran, Korvin, AnhNhan
Differential Revision: https://secure.phabricator.com/D5634
Summary:
Adds a policy-aware query class for selecting Releeph projects. This doesn't really change anything.
- Make `ReleephProject` implment `PhabricatorPolicyInterface`, beginning the long journey to make it policy-aware.
- Implement `ReleephProjectQuery`, for querying projects using cursor-based, policy-aware paging.
- Use it on the list view, so we load only ~100 projects instead of all of them.
- Tweaked some of the URI routing stuff to make it a little more consistent with common practices.
Ref T2714.
Test Plan:
{F36434}
{F36435}
Reviewers: edward
Reviewed By: edward
CC: aran
Maniphest Tasks: T2714
Differential Revision: https://secure.phabricator.com/D5390
