Summary:
See brief discussion in D18554. All the index tables are the same for every application (and, at this point, seem unlikely to change) and we never actually pass these objects around (they're only used internally).
In some other cases (like Transactions) not every application has the same tables (for example, Differential has extra field for inline comments), and/or we pass the objects around (lots of stuff uses `$xactions` directly).
However, in this case, and in Edges, we don't interact with any representation of the database state directly in much of the code, and it doesn't change from application to application.
Just automatically define document, field, and ngram tables for anything which implements `FerretInterface`. This makes the query and index logic a tiny bit messier but lets us delete a ton of boilerplate classes.
Test Plan: Indexed objects, searched for objects. Same results as before with much less code. Ran `bin/storage upgrade`, got a clean bill of health.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T12819
Differential Revision: https://secure.phabricator.com/D18559
Summary:
Ref T12819. I started trying to get individual engines to drive these constraints (e.g., `ManiphestTaskQuery` can do most of the work) but this is a big pain, especially since most engines don't support "any owner" or "no owner", and not everything has an owner, and so on and so on. Going down this path would have meant a huge pile of stub functions everywhere, I think.
Instead, drive these through the main engine using the fulltext document table, which already has everything we need to apply these constraints in a uniform way.
Also tweak some parts of query construction and result ordering.
Test Plan: Searched for documents by author, owner, unowned, any owner, tags, subscribers, fulltext in global search. Got sensible results without any application-specific code.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T12819
Differential Revision: https://secure.phabricator.com/D18550
Summary:
Ref T12819. Provides a Ferret-engine-based fulltext engine to ultimately replace the InnoDB fulltext engine.
This is still pretty basic (hard-coded and buggy) but technically sort of works.
To activate this, you must explicitly configure it, so it isn't visible to users yet.
Test Plan: Searched for objects with global fulltext search, got a mixture of matching revisions and tasks back.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T12819
Differential Revision: https://secure.phabricator.com/D18548
Summary: Ref T12819. Adds storage and indexing for the Ferret engine to Differential.
Test Plan: Ran `bin/search index D123 --force`, saw indexes appear in database. No UI/user impact yet.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T12819
Differential Revision: https://secure.phabricator.com/D18540
Summary:
Ref T2543. These are currently numeric values, like "0" and "3". I want to replace them with strings, like "accepted", and move definitions from Arcanist to Phabricator.
To set the stage for this, reduce the number of callsites where Phabricator invokes `ArcanistDifferentialRevisionStatus`.
This is just the easy ones. I'll hold this until the release cut.
Test Plan:
- Called `differential.find`.
- Called `differential.getrevision`.
- Called `differential.query`.
- Removed all reviewers from a revision, saw warning.
- Abandoned the no-reviewers revision, no more warning.
- Attached a revision to a task to get it to show the state icon with the status on a tooltip.
- Viewed revision bucketing on dashboard.
- Used `bin/search index` to reindex a revision.
- Hit the "Land Revision" endpoint.
I didn't explicitly test these cases:
- Doorkeeper Asana integration, since setup takes a thousand years.
- Disambiguation logic when multiple hashes match, since setup is also very involved.
- Releeph because it's Releeph.
Reviewers: chad
Reviewed By: chad
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18339
Summary:
Ref T10967. Improves some method names:
- `Revision->getReviewerStatus()` -> `Revision->getReviewers()`
- `Revision->attachReviewerStatus()` -> `Revision->attachReviewers()`
- `Reviewer->getStatus()` -> `Reviewer->getReviewerStatus()` (this is mostly to make this more greppable)
Test Plan:
- bunch o' `grep`
- Browsed around.
- If I missed anything, it should fatal in an obvious way. We have a lot of other `getStatus()` calls and it's hard to be sure I got them all.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10967
Differential Revision: https://secure.phabricator.com/D17522
Summary: Ref T10967. The old name was because we had a `getReviewers()` tied to `needRelationships()`, rename this method to use a simpler and more clear name.
Test Plan: `grep`, browsed around.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10967
Differential Revision: https://secure.phabricator.com/D17519
Summary: Ref T9979. This simplifies/standardizes the code a bit, but mostly gives us more consistent class names and structure.
Test Plan:
- Used `bin/search index --type ...` to index documents of every indexable type.
- Searched for documents by unique text, found them.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9979
Differential Revision: https://secure.phabricator.com/D14842
Summary:
Ref T9979. This is currently hard-coded but can be done in a generic way.
This has one minor behavioral changes: answer text is no longer included in the question text index in Ponder. I'm not planning to accommodate that for now since I don't want to dig this hole any deeper than I already have. This behavior should be different anyway (e.g., index the answer, then show the question in the results or something).
Test Plan:
- Put a unique word in a Maniphest comment.
- Searched for the word.
- Found the task.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9979
Differential Revision: https://secure.phabricator.com/D14837
Summary: Ref T5655. Rename `PhabricatorPHIDType` subclasses for clarity (see discussion in D9839). I'm not too keen on some of the resulting class names, so feel free to suggest alternatives.
Test Plan: Ran unit tests.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin, hach-que
Maniphest Tasks: T5655
Differential Revision: https://secure.phabricator.com/D9986
Summary:
Ref T2222. Ref T3886. Ref T418. A few changes:
- CustomField can now index into global search.
- Use CustomField fields instead of older custom fields for Differential global search. (This slightly breaks any custom fields which exist, but they are presumably very rare, and probably do not exist; this break is also very mild.)
- Automatically perform CustomField and Subscribable indexing on applicable object types.
Test Plan: Used `bin/search index` to reindex a bunch of stuff, then searched for it. Debug-dumped abstract documents to inspect them.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T418, T3886, T2222
Differential Revision: https://secure.phabricator.com/D8346
Summary:
Ref T2222. This is the big one.
This migrates each `DifferentialComment` to one or more ApplicationTransactions (action, cc, reviewers, update, comment, inlines), and makes `DifferentialComment` a double-reader for ApplicationTransactions.
The migration is pretty straightforward:
- If a comment took an action not otherwise covered, it gets an "action" transaction. This is something like "epriestley abandoned this revision.".
- If a comment updated the diff, it gets an "updated diff" transaction. Very old transactions of this type may not have a diff ID (probably only at Facebook).
- If a comment added or removed reviewers, it gets a "changed reviewers" transaction.
- If a comment added CCs, it gets a "subscribers" transaction.
- If a comment added comment text, it gets a "comment" transaction.
- For each inline attached to a comment, we generate an "inline" transaction.
Most comments generate a small number of transactions, but a few generate a significant number.
At HEAD, the code is basically already doing this, so comments in the last day or two already obey these rules, roughly, and will all generate only one transaction (except inlines).
Because we've already preallocated PHIDs in the comment text table, we only need to write to the transaction table.
NOTE: This significantly degrades Differential, making inline comments pretty much useless (they each get their own transaction, and don't show line numbers or files). The data is all fine, but the UI is garbage now. This needs to be fixed before we can deploy this to users, but it's easily separable since it's all just display code.
Specifically, they look like this:
{F112270}
Test Plan:
I've migrated locally and put things through their paces, but it's hard to catch sketchy stuff locally because most of my test data is nonsense and bad migrations wouldn't necessarily look out of place.
IMPORTANT: I'm planning to push this to a branch and then shift production over to the branch, and run it for a day or two before bringing it to master.
I generally feel good about this change: it's not that big since we were able to separate a lot of pieces out of it, and it's pretty straightforward. That said, it's still one of the most scary/dangerous changes we've ever made.
Reviewers: btrahan
CC: chad, aran
Maniphest Tasks: T2222
Differential Revision: https://secure.phabricator.com/D8210
Summary: Ref T4365. Two diffs from now, I'm changing the UI a bit to let you search for closed and unowned documents more explcitly. To support this in ElasticSearch and more easily in MySQL search, make these explicit, positive relationships.
Test Plan: `bin/search index --all`
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T4365
Differential Revision: https://secure.phabricator.com/D8122
Summary:
Currently, "Closed" and "Abandoned" are treated as "closed". I want to add a flag which treats "Accepted" as "Closed", too, for Asana and other companies who use an Asana-like workflow.
The background here is that their workflow is a bit weird. They basically do audits, but have a lot of things which Diffusion doesn't do well right now. This one change makes Differential fit their workflow fairly well, even though it's an audit workflow.
To prepare for this, normalize the definition of "closed" better. We have a few callsites which explicitly check for "ABANDONED || CLOSED", and normalizing this is cleaner anyway.
Also delete the very old COMMITTED status, which has been obsolete for over a year.
Test Plan: Browsed around most/all of the affected interfaces.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Differential Revision: https://secure.phabricator.com/D7653
Summary: Noticed this in the schema. "Touches" were an idea that never really got off the ground, as we built out more/better notification channels instead. Essentially, they recorded any object you'd ever interacted with. Maybe this will be useful some day, but for now it does nothing and can't be interacted with. Nuke it.
Test Plan: `grep`, loaded Maniphest.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Differential Revision: https://secure.phabricator.com/D6953
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 T2715. Had to start loading status information in the query class. Debated trying to clean up some of the attach / load stuff but decided to just add status under the new paradigm for now.
Test Plan: phid.query also made a status and checked that out. also played in conpherence.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Maniphest Tasks: T2715
Differential Revision: https://secure.phabricator.com/D6585
Summary:
Ref T2222. See D6260.
Push all this junk behind a Query so I can move the storage out from underneath it.
Test Plan: Viewed home page, list view, revision. Made draft, looked at preview, submitted draft, viewed inline, replied to inline.
Reviewers: btrahan
Reviewed By: btrahan
CC: chad, aran
Maniphest Tasks: T2222
Differential Revision: https://secure.phabricator.com/D6262
Summary:
Ref T2222.
I'm thinking about how I want to approach the Asana sync, and I want to try to do T2222 first so that we can build it cleanly on top of ApplicationTransactions. I think we can at least walk down this road a little bit and if it turns out to be scary we can take another approach.
I was generally very happy with how the auth migration turned out (seemingly, it was almost completely clean), and want to pursue a similar strategy here. Basically:
- Wrap the new objects in the old objects for reads/writes.
- Migrate all the existing data to the new table.
- Everything hard is done; move things over a piece at a time at a leisurely pace in lots of smallish, relatively-easy-to-understand changes.
This deletes or abstracts all reads of the DifferentialComment table. In particular, these things are **deleted**:
- The script `undo_commits.php`, which I haven't pointed anyone at in a very long time.
- The `differential.getrevisionfeedback` Conduit method, which has been marked deprecated for a year or more.
- The `/stats/` interface in Differential, which should be rebuilt on Fact and has never been exposed in the UI. It does a ton of joins and such which are prohibitively difficult to migrate.
This leaves a small number of reading interfaces, which I replaced with a new `DifferentialCommentQuery`. Some future change will make this actually load transactions and wrap them with DifferentialComment interfaces.
Test Plan: Viewed a revision; made revision comments
Reviewers: btrahan
Reviewed By: btrahan
CC: edward, chad, aran
Maniphest Tasks: T2222
Differential Revision: https://secure.phabricator.com/D6260
Summary:
Unmuck almost all of the we-sort-of-have-viewers-some-of-the-time mess.
There are a few notable cases here:
- I used Omnipotent users when indexing objects for search. I think this is correct; we do policy filtering when showing results.
- I cheated in a bad way in the Remarkup object rule, but fixing this requires fixing all the PhabricatorRemarkupEngine callsites (there are 85). I'll do that in the next diff.
- I cheated in a few random places, like when sending mail about package edits. These aren't a big deal.
Test Plan:
- Grepped for all PhabricatorObjectHandleData references.
- Gave them viewers.
Reviewers: vrana
Reviewed By: vrana
CC: aran, edward
Maniphest Tasks: T603
Differential Revision: https://secure.phabricator.com/D5151
Summary: Mostly applies a new call spacing rule; also a few things that have slipped through via pull requests and such
Test Plan: `find src/ -type f -name '*.php' | xargs -n16 arc lint --output summary --apply-patches`
Reviewers: chad
Reviewed By: chad
CC: aran
Differential Revision: https://secure.phabricator.com/D5002
Summary:
The search indexing API has several problems right now:
- Always runs in-process.
- It would be nice to push this into the task queue for performance. However, the API currently passses an object all the way through (and some indexers depend on preloaded object attributes), so it can't be dumped into the task queue at any stage since we can't serialize it.
- Being able to use the task queue will also make rebuilding indexes faster.
- Instead, make the API phid-oriented.
- No uniform indexing API.
- Each "Editor" currently calls SomeCustomIndexer::indexThing(). This won't work with AbstractTransactions. The API is also just weird.
- Instead, provide a uniform API.
- No uniform CLI.
- We have `scripts/search/reindex_everything.php`, but it doesn't actually index everything. Each new document type needs to be separately added to it, leading to stuff like D3839. Third-party applications can't provide indexers.
- Instead, let indexers expose documents for indexing.
- Not application-oriented.
- All the indexers live in search/ right now, which isn't the right organization in an application-orietned view of the world.
- Instead, move indexers to applications and load them with SymbolLoader.
Test Plan:
- `bin/search index`
- Indexed one revision, one task.
- Indexed `--type TASK`, `--type DREV`, etc., for all types.
- Indexed `--all`.
- Added the word "saboteur" to a revision, task, wiki page, and question and then searched for it.
- Creating users is a pain; searched for a user after indexing.
- Creating commits is a pain; searched for a commit after indexing.
- Mocks aren't currently loadable in the result view, so their indexing is moot.
Reviewers: btrahan, vrana
Reviewed By: btrahan
CC: 20after4, aran
Maniphest Tasks: T1991, T2104
Differential Revision: https://secure.phabricator.com/D4261
