Summary:
Depends on D18856. Ref T13036. See PHI275. When we receive a length frame but the buffer doesn't have any data yet, we currently emit a pointless 0-length data frame on the channel.
For normal chatter this is harmless/valid, but it causes problems when a channel has transitioned into bundle2 mode (probably it indicates "end of stream")?
In any case, it's never helpful, so if we're about to read a data block and don't have any data, just bail out until we see some more data.
Note that we can't end up here //expecting// a 0-length data block: both the `data-length` and `data-bytes` states already handle that properly.
Test Plan: Pushed 4MB changes to a Mercurial repository with Mercurial 4.1.1, was no longer able to hit channel errors.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13036
Differential Revision: https://secure.phabricator.com/D18857
Summary:
Depends on D18855. Ref T13036. This comment no longer seems to be accurate: anything we send over `stderr` is faithfully shown to the user with recent clients.
From [[ https://www.mercurial-scm.org/repo/hg/file/default/mercurial/help/internals/wireprotocol.txt | this document ]], the missing sauce may have been:
```
A generic error response type is also supported. It consists of a an error
message written to ``stderr`` followed by ``\n-\n``. In addition, ``\n`` is
written to ``stdout``.
```
That is, writing "\n" to stdout in addition to writing the error to stderr. However, this no longer appears to be necessary.
I think the modern client behavior is generally sensible (and consistent with the behavior of Git and Subversion) so this //probably// isn't a bug or me making a mistake.
Test Plan: With a modern client, threw some arbitrary exception during execution. Observed a helpful message on the client with no additional steps.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13036
Differential Revision: https://secure.phabricator.com/D18856
Summary:
Ref T13036. This code attempts to filter the "capabilities" message to remove "bundle2", but I think this has never worked.
Specifically, the //write// pathway is hooked, and "write" here means "client is writing a message to the server". However, the "capabilities" frame is part of the response, not part of the request. Thus, this code never fires, at least on recent versions of Mercurial.
Since I plan to support bundle2 and don't want to decode response frames, just get rid of this, assuming we'll achieve those goals.
I think this was just overlooked in D14241, which probably focused on the HTTP version. This code does (at least, potentially) do something for HTTP.
I'm leaving the actual "strip stuff" code in place for now since I think it's still used on the HTTP pathway.
Test Plan:
- Added debug logging, saw this code never hit even though `hg push --debug` shows the client believing bundle2 is supported.
- Logged both halves of the wire protocol and saw this come from the server, not the client.
- Ran the failing `hg push` of a 4MB file under hg 4.4.1, got the same error as before.
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: cspeckmim
Maniphest Tasks: T13036
Differential Revision: https://secure.phabricator.com/D18855
Summary:
Depends on D18848. Ref PHI243. This puts a bit of logic up front to figure out the blueprint type before we actually start editing it.
This implementation is a little messy but it keeps the API clean. Eventually, the implementation could probably go in the TransactionTypes so more code is shared, but I'd like to wait for a couple more of these first.
This capability probably isn't too useful, but just pays down a bit of technical debt from the caveat introduced in D18822.
Test Plan:
- Created a new blueprint with the API.
- Tried to create a blueprint without a "type" (got a helpful error).
- Created and edited blueprints via the web UI.
- Tried to change the "type" of an existing blueprint (got a helpful error).
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Differential Revision: https://secure.phabricator.com/D18849
Summary:
Depends on D18845. See PHI243 for context and more details.
Briefly, some objects need a "type" transaction (or something similar) very early on, and we can't generate their fields until we know the object type. Drydock blueprints are an example: a blueprint's fields depend on the blueprint's type.
In web interfaces, the workflow just forces the user to select a type first. For Conduit workflows, I think the cleanest approach is to proactively extract and apply type information before processing the request. This will make the implementation a little messier, but the API cleaner.
An alternative is to add more fields to the API, like a "type" field. This makes the implementation cleaner, but the API messier. I think we're better off favoring a cleaner API here.
This change just makes it possible for `DrydockBlueprintEditEngine` to look at the incoming transactions and extract a "type"; it doesn't actually change any behavior.
Test Plan: Performed edits via API, but this change doesn't alter any behavior.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18847
Summary: Ref PHI243. This is a followup to D18822, which added an edit-only `drydock.blueprint.edit`. By modularizing transactions (here) and then adding a "type" transaction (next change) I intend to remove the "edit-only" limitation and make this API method fully functional.
Test Plan: Created and edited blueprints via the web UI. Edited blueprints via the API. Disabled/enabled blueprints (currently web UI only).
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Differential Revision: https://secure.phabricator.com/D18845
Summary:
Fixes T13031. "Enormous" changes are basically changes which are too large to hold in memory, although the actual definition we use today is "more than 1GB of change text or `git diff` runs for more than 15 minutes".
If an install configures a Herald content rule like "when content matches /XYZ/, do something" and then a user pushes a 30 GB source file, we can't put it into memory to `preg_match()` it. Currently, the way to handle this case is to write a separate Herald rule that rejects enormous changes. However, this isn't obvious and means the default behavior is unsafe.
Make the default behavior safe by rejecting these changes with a message, similar to how we reject "dangerous" changes (which permanently delete or overwrite history) by default.
Also, change a couple of UI strings from "Enormous" to "Very Large" to reduce ambiguity. See <https://discourse.phabricator-community.org/t/herald-enormous-check/822>.
Test Plan: Changed the definition of "enormous" from 1GB to 1 byte. Pushed a change; got rejected. Allowed enormous changes, pushed, got rejected by a Herald rule. Disabled the Herald rule, pushed, got a clean push. Prevented enormous changes again. Grepped for "enormous" elsewhere in the UI.
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: joshuaspence
Maniphest Tasks: T13031
Differential Revision: https://secure.phabricator.com/D18850
Summary:
See PHI273. Ref T13020. After D18777, tasks created directly into the default status (which is common) via the web UI no longer write a "status" transaction.
This is consistent with other applications, and consistent with the API/email behavior for tasks since early 2016. It also improves the consistency of //reading// tasks via the API.
However, it impacted the "Burnup Report" which relies on directly reading these rows to detect task creation. Until this is fixed properly (T1562), synthetically generate the "missing" transactions which this page expects by looking at task creation dates instead.
Specifically, we:
- Generate a fake `status: null -> "open"` transaction for every task by looking at the Task table.
- Go through the transaction list and remove all the legacy `status: null -> "any open status"` transactions. These will only exist for older tasks.
- Merge all our new fake transactions into the list of transactions.
- Continue on as though nothing happened, letting the rendering code continue to operate on legacy-looking data.
I think this will slightly miscount tasks which were created directly into a closed status, but this is very rare, and does not significantly impact the accuracy of this report relative to other known issues (notably, merging closed tasks).
This will also get the wrong result if the default status has changed from an "open" status to a "closed" status at any point, but this is exceptionally bizarre/rare.
Ultimately, T1562 will let us delete all this stuff and disavow its existence.
Test Plan:
- Created some tasks, loaded burnup before/after this patch.
- My local chart looks more accurate afterwards, but the data is super weird (I used `bin/lipsum` to create a huge number of tasks a couple months ago). I'll vet this on `secure`, which has more reasonable data.
Here's my local chart:
{F5356499}
That's what it //should// look like, it's just hard to be confident that nothing else is hiding there.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13020
Differential Revision: https://secure.phabricator.com/D18853
Summary:
Depends on D18851. Ref T13035. After D18819, revision creation transactions may be split into two groups (if prototypes are enabled).
This split means we have two workers. The first worker doesn't publish feed stories or mail; the second one does.
Currently, both workers call `shouldPublishFeedStory()` before they queue, and then again after the daemons pull them out of the queue. However, the answer to this question can change.
Specifically, this happens:
- `arc` creates a revision.
- The first transaction group applies, creating the revision as a draft, and returns `false` from `shouldPublishFeedStory()`, and does not generate related PHIDs. It queues a daemon to send mail, expecting it not to publish a feed story.
- The second transaction group applies, promoting the revision to "needs review". Since the revision has promoted, `shouldPublishFeedStory()` now returns true. This editor generates related PHIDs and queues a daemon task, expecting it to send mail / publish feed.
- A few milliseconds pass.
- The first job gets pulled out of the daemon queue and worked on. It does not have any feed metadata because the object wasn't publishable when the job was queued -- but `shouldPublishFeedStory()` now returns true, so it tries to publish a story without any metadata available. Slightly bad stuff happens (see below).
- The second job gets pulled out of the daemon queue and worked on. This one has metadata and works fine.
The "slightly bad stuff" is that we publish an empty feed story with no references to any objects, then try to push it to hooks and other listeners. Since it doesn't have any references, it fails to load during the "push to external listeners" phase.
This is harmless but clutters the log and doesn't help anything.
Instead, cache the state of "are we publishing a feed story for this object?" when we queue the worker so it can't race.
Test Plan:
- Enabled prototypes.
- Disabled all Herald triggers for Harbormaster build plans.
- Ran `bin/phd debug task` in one window.
- Created a revision in a separate window.
- Before patch: saw "unable to load feed story" errors in the daemon log.
- After patch: no more "unable to load feed story" errors.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13035
Differential Revision: https://secure.phabricator.com/D18852
Summary: Ref T13035. See that task for a description of the issue.
Test Plan:
- Enabled prototypes.
- Disabled all Herald rules that trigger Harbormaster builds.
- Created a new revision.
- Before patch: initial review request email was dropped.
- After patch: initial review request email is sent.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13035
Differential Revision: https://secure.phabricator.com/D18851
Summary:
See PHI262. Fixes T12578. Although this is a bit niche and probably better accomplished through advisory/soft measures ("Add blocking reviewers") in most cases, it isn't difficult to implement and doesn't create any technical or product tension.
If installs write a rule that blocks commits, that will probably also naturally lead them to an "add reviewers" rule anyway.
Also, allow packages to be hit with the typeahead. They're valid reviewers but previously you couldn't write rules against them, for no actual reason.
Test Plan: Used test console to run this against commits, got sensible results for the field value.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12578
Differential Revision: https://secure.phabricator.com/D18839
Summary:
See PHI223. Ref T13024. There's a remaining registration/login order issue after the other changes in T13024: we lose track of the current URI when we go through the MFA flow, so we can lose "Set Password" at the end of the flow.
Specifically, the flow goes like this today:
- User clicks the welcome link in email.
- They get redirected to the "set password" settings panel.
- This gets pre-empted by Legalpad (although we'll potentially survive this with the URI intact).
- This also gets pre-empted by the "Set MFA" workflow. If the user completes this flow, they get redirected to a `/auth/multifactor/?id=123` sort of URI to highlight the factor they added. This causes us to lose the `/settings/panel/password/blah/blah?key=xyz` URI.
The ordering on this is also not ideal; it's preferable to start with a password, then do the other steps, so the user can return to the flow more easily if they are interrupted.
Resolve this by separating the "change your password" and "set/reset your password" flows onto two different pages. This copy/pastes a bit of code, but both flows end up simpler so it feels reasonable to me overall.
We don't require a full session for "set/reset password" (so you can do it if you don't have MFA/legalpad yet) and do it first.
This works better and is broadly simpler for users.
Test Plan:
- Required MFA + legalpad, invited a user via email, registered.
- Before: password set flow got lost when setting MFA.
- After: prompted to set password, then sign documents, then set up MFA.
- Reset password (with MFA confgiured, was required to MFA first).
- Tried to reset password without a valid reset key, wasn't successful.
- Changed password using existing flow.
- Hit various (all?) error cases (short password, common password, mismatch, missing password, etc).
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024
Differential Revision: https://secure.phabricator.com/D18840
Summary:
Ref PHI261. This moves Harbormaster build status to work more similarly to other modern status types, like Differential revision status, where we try to specify as much behavior on the server as possible so that the client and server can vary independently.
(I don't have any specific plans to make Harbormaster build status configurable on the server side, but it isn't out of the question.)
Test Plan: Ran `harbormaster.querybuilds` (saw same data as before) and `harbormaster.build.search` (saw same data as before, plus new ANSI color data).
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Differential Revision: https://secure.phabricator.com/D18838
Summary: This probably isn't the best solution, however, it conveniently avoids the bug from T13033. It would probably be more user-friendly (but more difficult to implement) if we allowed either Project Details //or// Workboard to be hidden but not both.
Test Plan: Tested locally, indeed this prevents hiding the menu item.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D18836
Summary:
Fixes T13033. Currently (prior to D18836) all the default items for projects can be hidden. When this occurs, the main project page fatals.
If we fix the fatal narrowly (don't try to call `null->getBuiltinKey()` when `$default` is `null`), it would 404, which is a little better but not by much.
After D18836 you can't hide "Profile", which is pretty sensible, and effectively fixes this. This change doubles down: let "Manage" be a default, and send the user there if we can't find a different default.
Ideally, the MenuEngine itself will do more rendering eventually (as it does for some of the newer Home stuff) and could handle this defaulting behavior with less special casing, but we'd still end up in a similar situation if a project had only one "Link" item to "coolcats.com" or something: redirecting the user to "coolcats.com" is probably better than fataling, but not by a huge margin, and not likely to be what they expect.
Test Plan:
Before D18836, disabled both "Profile" and "Workboard" items on a project. Visited project page.
Before patch: fatal. After patch: manage page.
After D18836, you can't do this and just get the profile, so this is sort of moot and mostly future-proofing/for-completeness.
Reviewers: 20after4, amckinley
Reviewed By: amckinley
Maniphest Tasks: T13033
Differential Revision: https://secure.phabricator.com/D18843
Summary: Events with no attendees (e.g. fresh instances of recurring events) would trigger an exception when sending notifications, because `$attendee_map` would still get populated.
Test Plan: Declined event, restarted daemons. Did not see exception. Accepted event, restarted daemons. Saw "[Calendar] [Reminder]" email.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D18835
Summary:
We have one production instance with failing database backups since they recently uploaded a 52MB hunk. The production configuration specifies a 64MB "max_allowed_packet" in `[mysqld]`, but this doesn't apply to `mysqldump` (we'd need to specify it in a separate `[mysqldump]` section) and `mysqldump` runs with an effective limit of the default (16MB).
We could change our production config to specify a value in `[mysqldump]`, but just change it unconditionally at execution time since there's no reason for any user to ever want this command to fail because they have too much data.
Test Plan: Dumped locally, will verify production backup goes through cleanly.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18834
Summary: Ref T13030. See PHI254. This behavior could be cleaner than I've made it, but it fixes the "this is totally broken" issue, replacing a fatal/exception with an informative (just not terribly useful) page.
Test Plan:
- Added a submodule to a repository.
- In Diffusion, clicked some other file next to the submodule, then edited the URI to the submodule path instead.
- Before patch: fatal.
- After patch: relatively useful message about this being a submodule.
Note that it's normally hard to hit this URI directly. In the browse view, submodules are marked up as directories and linked to a separate submodule resolution flow.
{F5321524}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13030
Differential Revision: https://secure.phabricator.com/D18831
Summary:
See PHI230. Currently, we denormalize raw line counts onto diffs and revisions, but not added/removed line counts.
I'd like to try a `[---+ ]` sort of size hint element (see D16322 for more) as a general approach to conveying size information at a glance and see how it feels, since I think the raw size number isn't very scannable/useful and it may be a significant improvement to hint about how much of a change is throwing stuff out vs adding new stuff.
This just makes the data available without any subquerying and doesn't actually change the UI.
Test Plan:
Created a revision, saw detailed change information populate in the database.
```
mysql> select * from differential_revision where id = 292\G
*************************** 1. row ***************************
id: 292
title: WIP
originalTitle: WIP
phid: PHID-DREV-ux3cxptibn3l5pxsug3z
status: draft
summary: asdf
testPlan: asdf
authorPHID: PHID-USER-cvfydnwadpdj7vdon36z
lastReviewerPHID: NULL
lineCount: 41
dateCreated: 1513179418
dateModified: 1513179418
attached: []
mailKey: h4mn6perdio47o4beomyvu75zezwvredx3mbrlgz
branchName: NULL
viewPolicy: users
editPolicy: users
repositoryPHID: PHID-REPO-wif5lutk5gn3y6ursk4p
properties: {"lines.added":40,"lines.removed":1}
activeDiffPHID: PHID-DIFF-ixjphpunpkenqgukpmce
1 row in set (0.00 sec)
```
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18832
Summary:
Depends on D18828. Ref T7789. See <https://discourse.phabricator-community.org/t/git-lfs-fails-with-large-images/584>.
Currently, when you upload a large (>4MB) image, we may try to assess the dimensions for the image and for each individual chunk.
At best, this is slow and not useful. At worst, it fatals or consumes a ton of memory and I/O we don't need to be using.
Instead:
- Don't try to assess dimensions for chunked files.
- Don't try to assess dimensions for the chunks themselves.
- Squelch errors for bad data, etc., that `gd` can't actually read, since we recover sensibly.
Test Plan:
- Created a 2048x2048 PNG in Photoshop using the "Random Noise" filter which weighs 8.5MB.
- Uploaded it.
- Before patch: got complaints in log about `imagecreatefromstring()` failing, although the actual upload went OK in my environment.
- After patch: clean log, no attempt to detect the size of a big image.
- Also uploaded a small image, got dimensions detected properly still.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T7789
Differential Revision: https://secure.phabricator.com/D18830
Summary: Depends on D18827. Ref T7789. See PHI204. See PHI131. This button got accidentally removed in Diffusion refactoring (`$data` is no longer used).
Test Plan: {F5321459}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T7789
Differential Revision: https://secure.phabricator.com/D18828
Summary: See PHI131. Ref T7789. Although this probably isn't 100% complete, there don't seem to be any actual, known, practical blocking issues remaining (everything is either heresay or not reproducible).
Test Plan: Tried to push LFS locally, got blocked with a helpful message. Enabled setting, tried to push LFS locally, got a successful push.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T7789
Differential Revision: https://secure.phabricator.com/D18825
Summary:
See PHI242. All use cases for this that I know of are pretty hacky, but they don't seem perilous, and it's easier than webhooks.
See P1895, T10183, and T9853 for me previously refusing to implement this since all those use cases were also pretty bad.
Test Plan:
- Wrote a rule to add comments, saw it add comments.
- Reviewed summary, re-edited rule, reviewed transcript to check that all the strings worked OK.
- Wrote a new rule for a non-commentable object (a blog) to make sure I wasn't offered the "Add a comment" action.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18823
Summary:
Ref: https://admin.phacility.com/PHI243
Since our use case primarily focuses on transaction editing, this patch implements the `drydock.blueprint.edit` api method with the understanding that:
a) this is a work in progress
b) object editing is supported, but object creation is not yet implemented
Test Plan:
* updated existing blueprints via Conduit UI
* regression tested `maniphest.edit` by creating new and updating existing tasks
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin, yelirekim, jcox
Differential Revision: https://secure.phabricator.com/D18822
Summary:
Fixes T13027. Ref T2543. When revisions promote from "Draft" because builds finish or no builds are configured, the status currently switches from "Draft" to "Needs Review" without re-running Herald.
This means that some rules -- notably, "Send me an email" rules -- don't fire as soon as they should.
Instead of applying this promotion in a hacky way inline, queue it and apply it normally in a second edit, after the current group finishes.
Test Plan:
- Created a revision, reviewed Herald transcripts.
- Saw three Herald passes:
- First pass (revision creation) triggered builds and no email.
- Second pass (builds finished) did not trigger builds (no update) and did not trigger email (revision still a draft).
- Third pass (after promotion out of 'draft') did not trigger builds (no update) but did trigger email (revision no longer a draft).
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13027, T2543
Differential Revision: https://secure.phabricator.com/D18819
Summary:
See <https://discourse.phabricator-community.org/t/activation-link-in-welcome-mail-only-works-if-new-user-isnt-semi-logged-in/740/7>.
In T13024, I rewrote the main menu bar to hide potentially sensitive items (like notification and message counts and saved search filters) until users fully log in.
However, the "Log In" item got caught in this too. For clarity, rename `shouldAllowPartialSessions()` to `shouldRequireFullSession()` (since logged-out users don't have any session at all, so it would be a bit misleading to say that "Log In" "allows" a partial session). Then let "Log In" work again for logged-out users.
(In most cases, users are prompted to log in when they take an action which requires them to be logged in -- like creating or editing an object, or adding comments -- so this item doesn't really need to exist. However, it aligns better with user expectations in many cases to have it present, and some reasonable operations like "Check if I have notifications/messages" don't have an obvious thing to click otherwise.)
Test Plan: Viewed site in an incognito window, saw "Log In" button again. Browsed normally, saw normal menu.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18818
Summary:
See <https://discourse.phabricator-community.org/t/diffusion-observed-mercurial-repository-history-broken/825>.
In D18769, I rewrote this from using the `--branch` flag (which is unsafe and does not function on branches named `--config=x.y` and such).
However, this rewrite accidentally changed the result order, which impacted Mercurial commit hisotry lists and graphs. Swap the order of the constraints so we get newest-to-oldest again, as expected.
Test Plan: Viewed a Mercurial repository's history graph, saw sensible chronology after the patch.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18817
Summary: Ref T13001, URLs that return multiple commits should show a list of those commits. Not sure if the actual list looks very pretty this way, but was wondering if this approach was vaguely correct.
Test Plan:
- Navigate to `install/rPbd3c23`
- User should see a list view providing links to `install/rPbd3c2355e8e2b220ae5e3cbfe4a057c8088c6a38` and `install/rPbd3c239d5aada68a31db5742bbb8ec099074a561`
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin
Maniphest Tasks: T13001
Differential Revision: https://secure.phabricator.com/D18816
See PHI238. When an install uninstalls "Legalpad", we were incorrectly failing
to mark sessions as "Signed All Required Documents" by bailing early.
Test Plan: Uninstalled Legalpad, logged in.
Summary: Ref T13019, adds build status back to Diffusion commits
Test Plan: Open a Diffusion commit that has a build status, property list view should show the build status, but not Subscriptions, Projects, or Tokens.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin
Maniphest Tasks: T13019
Differential Revision: https://secure.phabricator.com/D18813
Summary: See PHI234. In T12931 we improved the behavior of Diffusion when a repository's default branch is set to a branch that does not exist, but in T11823 the way refcursors work changed, and we can now get a cursor (just with no positions) back for a deleted branch. When we did, we didn't handle things gracefully.
Test Plan:
- Set default branch to a deleted branch, saw nice error instead of fatal.
- Set default branch to a nonexistent branch which never existed, saw nice error.
- Set default branch to existing "master", saw repository normally.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18811
Summary:
See PHI234. Several issues here:
- The warning about observing a repository in Read/Write mode checks the raw I/O type, not the effective I/O type. That means we can fail to warn if other URIs are set to "Default", and "Default" is "Read/Write" in practice.
- There's just an actual typo which prevents the "Observe" version of this error from triggering properly.
Additionally, add more forceful warnings that "Observe" and "Mirror" mean that you want to //replace// a repository with another one, not that we somehow merge branches selectively. It isn't necessarily obvious that "Observe" doesn't mean "merge/union", since the reasons it can't in the general case are somewhat subtle (conflicts between refs with the same names, detecting ref deletion).
Test Plan:
Read documentation. Hit the error locally by trying to "Observe" while in Read/Write mode:
{F5302655}
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18810
Summary:
Ref T10405. If `gd` is not installed, a number of unit tests currently fail because they generate synthetic users and try to composite profile pictures for them.
Instead, just fall back to a static default if `gd` is not available.
Test Plan: Faked this pathway, created a new user, saw a default profile picture.
Reviewers: amckinley, lpriestley, avivey
Reviewed By: avivey
Maniphest Tasks: T10405
Differential Revision: https://secure.phabricator.com/D18812
Summary:
Ref T10233. See PHI231. When users ignore the `arc land` prompt about bad revision states, make it explicitly clear in the transaction log that they broke the rules.
You can currently figure this out by noticing that there's no "This revision is accepted and ready to land." message, but it's unrealistic to expect non-expert users to look for the //absence// of a message to indicate something, and this state change is often relevant.
Test Plan: {F5302351}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T10233
Differential Revision: https://secure.phabricator.com/D18808
Summary:
Use ClassQuery to find datasources for the quick-search.
Mostly, this allows extensions to add quicksearches.
Test Plan:
using `/typeahead/class/`, tested several search terms that make sense.
Removed the tag interface from a datasource, which removed it from results.
Reviewers: epriestley, amckinley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D18760
Summary:
Depends on D18801. Ref T2543. See PHI229. I missed "Accept" before, but intended to disallow it (like "Reject") since I don't want drafts to be reviewable.
However, "Resign" seems fine to allow? So let's allow that for now.
Test Plan: Was no longer offered "Accept" on draft revisions.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18802
Summary:
See PHI228. Ref T2543. The current logic gets this slightly wrong: prototypes are off, you create a draft with `--draft`, then promote it with "Request Review". This misses both branches.
Instead, test these conditions a little more broadly. We also need to store broadcast state since `getIsNewObject()` isn't good enough with this workflow.
Test Plan:
- With prototypes on and autopromotion, got a rich email after builds finished.
- With prototypes off, got a rich email immediately.
- With prototypes off and `--draft`, got a rich email after "Request Review".
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18801
Summary:
See PHI173. Adds custom field support for Herald actions, and implements actions for "Datasource/Tokenizer" fields.
The only action available for now is "set field to...". Other actions ("Add values", "Remove values") might make sense in the future for these fields, but there's currently no use case. For most other field types (text, select, checkbox, etc) only "Set to" makes sense.
Test Plan:
- Added a "datasource" custom field to the custom field definition in Config.
- Added a "if field is empty, set field to default value X" rule to Herald.
- Created a task with a nonempty field: no Herald trigger.
- Created a task with an empty field: Herald fired.
- Reviewed rule and transcripts for text strings.
{F5297615}
{F5297616}
{F5297617}
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18784
Summary:
Depends on D18792. Fixes T13024. Fixes T89198. Currently, when users are logging in initially (for example, need to enter MFA) we show more menu items than we should.
Notably, we may show some personalized/private account details, like the number of unread notifications (probably not relevant) or a user's saved queries (possibly sensitive). At best these are misleading (they won't work yet) and there's an outside possibility they leak a little bit of private data.
Instead, nuke everything except "Log Out" when users have partial sessions.
Test Plan:
Hit a partial session (MFA required, email verification required) and looked at the menu. Only saw "Log Out".
{F5297713}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024
Differential Revision: https://secure.phabricator.com/D18793
Summary: Depends on D18791. Ref T13024. This clears up another initialization order issue, where an unverified address could prevent MFA enrollment.
Test Plan: Configured both verification required and MFA required, clicked "Add Factor", got a dialog for the workflow.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024
Differential Revision: https://secure.phabricator.com/D18792
Summary:
Depends on D18790. Ref T13024. Fixes T8335. Currently, "unapproved" and "disabled" users are bundled together. This prevents users from completing some registration steps (verification, legalpad documents, MFA enrollment) before approval.
Separate approval out and move it to the end so users can do all the required enrollment stuff on their end before we roadblock them.
Test Plan: Required approval, email verification, signatures, and MFA. Registered an account. Verified email, signed documents, enrolled in MFA, and then got prompted to wait for approval.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024, T8335
Differential Revision: https://secure.phabricator.com/D18791
Summary:
Depends on D18789. Ref T13024. See PHI223. Currently, if `security.require-multi-factor-auth` and Legalpad "Signature Required" documents are //both// set, it's not possible to survive account registration, since MFA is requiried to sign and signatures are required to add MFA.
Instead, check for signatures before requiring MFA enrollment. This makes logical sense, since it's silly to add MFA if you don't agree to a Terms of Service or whatever.
(Note that if you already have MFA, we prompt for that first, before either of these steps, which also makes sense.)
Test Plan: Configured `security.require-multi-factor-auth`. Added a signature-required document. Loaded a page as a new user. Went through signature workflow, then through the MFA enrollment workflow.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024
Differential Revision: https://secure.phabricator.com/D18790
Summary: Depends on D18788. Ref T13024. Currently, we prompt users to sign from newest to oldest. This seems less intuitive than oldest to newest.
Test Plan: Dumped document order, saw it swap to oldest-first.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024
Differential Revision: https://secure.phabricator.com/D18789
Summary: Depends on D18786. Ref T13024. I'm going to change the order this occurs in, but move it to a separate method and clean it up a little first.
Test Plan: Added a new document as required, reloaded, signed it, got logged into a session.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024
Differential Revision: https://secure.phabricator.com/D18788
Summary: Depends on D18785. Ref T13024. While I'm in here, update this a bit to use the newer stuff.
Test Plan: Searched for Legalpad documents, saw more modern support (subscribers, order by, viewer()).
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024
Differential Revision: https://secure.phabricator.com/D18786
Summary: Ref T13024. Updates LegalpadDocumentQuery to use slightly more modern constructions.
Test Plan: Queried for Legalpad documents, got the same results.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13024
Differential Revision: https://secure.phabricator.com/D18785
Summary:
See PHI225. Previously, see D15335 / T10413. On workboards, we hide archived project tags since they aren't terribly useful in that context, at least most of the time. Originally, see T10349#159916 and D15297.
However, hovercards reuse this display logic, and it's inconsistent/confusing to hide them there, since the actual "Tags" elements on task pages show them. Narrow the scope of this rule.
Test Plan:
- Viewed a hovercard for a task with an archived project tagged, saw archived project.
- Viewed a workboard for the same task, saw only unarchived projects other than the current board tagged (this behavior is unchanged).
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18783
Summary:
Ref T13018. See that task and the Discourse thread for discussion.
This doesn't work as-is and we need to `og:description` everything to make it work. I don't want to sink any more time into this so just back all the changes out for now.
(The `<html>` change is unnecessary anyway.)
Test Plan: Strict revert.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13018
Differential Revision: https://secure.phabricator.com/D18782
Summary: Ref T13018. This is easy to get working roughly, at least, and seems reasonable.
Test Plan: Viewed page source, saw tags. Custom header logo still worked. Pretty hard to debug against a local install since Disqus / debugger tools can't hit it, but I'll see what it looks like in production and tweak it if I got anything horribly wrong.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13018
Differential Revision: https://secure.phabricator.com/D18780
Summary:
Ref T13020. See PHI221.
Freeze legacy method `maniphest.gettasktransactions` in favor of modern method `transaction.search`.
Remove legacy "null on create" behavior from Maniphest status and priority transactions. This behavior is obsolete with EditEngine, and leads to inconsistent transaction sets in the transaction record.
The desired behavior is that transactions which don't do anything (e.g., default value was not changed) don't appear in the transaction log.
Test Plan:
- Viewed API UI and saw `maniphest.gettasktransactions` marked as "Frozen".
- Created a new task via web UI (without changing status/priority), queried transactions with `maniphest.gettasktransacitons`/`transaction.search`, no longer saw "null on create" no-op transactions in record.
- Web UI is unchanged, since these transactions were hidden before and now do not exist.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13020
Differential Revision: https://secure.phabricator.com/D18777
Summary: Whitespace has semantic meaning for yaml files, so we shouldn't suppress whitespace-only lines of diff by default.
Test Plan: Edited local config to include yaml files, saw expected whitespace changes.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D18775
Summary:
See PHI180. Currently, if you begin creating or editing an inline and then swap display modes (for example, with "View Unified"), your edit is lost.
Persisting the editor state is complicated and this is very rare, so just prevent the action and warn the user instead.
Also make the warning persist for a little longer since a few of the messages, including this one, take a couple seconds to read now.
Test Plan:
- Edited a comment, tried to swap display modes, got a warning.
- Swapped display modes normally with no comment being edited.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18774
Summary:
Fixes paging on the Diffusion Repository List.
PhabricatorRepositoryQuery needs to specify a behavior for `null` on the OderableColumns definition for the `callsign` column.
See https://phabricator.wikimedia.org/T180457
Test Plan:
1. On an instance with more than 100 repositories
* some of which are missing a callsign
2. Attempt to sort by callsign.
3. See the sorted results
Previously:
3. Exception: "Column "0" has null value, but does not specify a null behavior."
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D18773
Summary:
Depends on D18771. See PHI206. Currently, `arc diff --draft` only holds revisions in draft mode: it doesn't put them into draft mode if the install isn't configured to use draft mode.
Instead, make it a bit more forceful so that `arc diff --draft` can create into draft mode explicitly even if protoypes are off. This aligns with expection a little more clearly.
Test Plan: Ran `arc diff --draft` with prototypes off, got a revision held in draft mode.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18772
Summary: See PHI210. Ref T2543. Currently, we don't set this flag if you have prototypes off and don't get any of the new draft stuff, so the mail drops some of the details it is supposed to have.
Test Plan: Disabled prototypes, created a revision, saw summary / test plan in the initial mail.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18771
Summary:
Ref T13012. These flags can be exploited by attackers to execute code remotely. See T13012 for discussion and context.
Additionally, harden some Mercurial commands where possible (by using additional quoting or embedding arguments in other constructs) so they resist these flags and behave properly when passed arguments with these values.
Test Plan:
- Added unit tests.
- Verified "--config" and "--debugger" commands are rejected.
- Verified more commands now work properly even with branches and files named `--debugger`, although not all of them do.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13012
Differential Revision: https://secure.phabricator.com/D18769
Summary:
See PHI199. Ref T2543. When you run a RevisionQuery with a legacy status constraint (via `differential.query`), we currently don't match "Draft" revisions.
Use the actual complete map from `DifferentialRevisionStatus` instead of hard coding the status list so "Draft" is included.
Test Plan:
- Ran `differential.query` with `ids` and `status` for a draft revision.
- Before patch: revision not returned in results.
- After patch: revision returned in results.
(Note that it returns as "Needs Review", for compatibility.)
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18765
Summary:
See PHI190. This clarifies the ruleset a bit:
- If you accepted, then the author used "Request Review" explicitly, we now show "Accepted Earlier" instead of "Accepted" in the "Reviewers" list on the main revision page. This makes it sligthly more clear why the revision is back in your review queue without picking through the transaction log.
- Instead of moving all non-current accepts into "Ready to Review", move only voided accepts into "Ready to Review". This stops us from pulling older accepts which haven't been voided (which could have been incorrectly pulled) and correctly pulls older, voided accepts from before an update (for example: accept, then request review, then update) and generally aligns better with intent/expectation.
Test Plan:
- Accepted, requested review.
- Saw reviewer as "Accepted Earlier".
- Saw review in "Ready to Review" bucket.
- Accepted, updated (with sticky accept).
- Saw reviewer as "Accepted Prior Diff".
- Saw review as "Waiting on Authors".
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18764
Summary:
Ref PHI193. This method of enforcing policy checks is now (mostly) obsolete, and they're generally checked at the Controller/API level instead.
Notably, this method does not call `adjustObjectForPolicyChecks(...)` properly, so it can not handle special cases like "creating a project and taking its newly created members into account" for object policies like "Project Members".
Just remove these checks, which are redundant with checks elsewhere.
Test Plan:
- Set Project application default edit policy to "Administrators and Project Members".
- Tried to create a project as a non-administrator, adding myself.
- Before patch: policy fatal on a VOID object (the project with no PHID generated yet).
- After patch: object created properly. Got a sensible policy error if I didn't include myself as a member.
- Also verified that other edit rules are still enforced/respected (I can't edit stuff I shouldn't be able to edit).
- There's at least a bit of unit test coverage of this, too, which I updated to work via API (which hits the new broad capability checks) instead of via low-level transactions (which enforce only a subset of policy operations now).
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18763
Summary: See PHI195. This bulks out these API methods since all the requests are pretty straightforward.
Test Plan: Ran `edge.search` and `differential.revision.search`.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18762
Summary:
See PHI197. Populates "status" transactions and a few other obvious types where there's no security/performance/payload/formatting issue I can come up with.
The names here are the same as the names for editing with `maniphest.edit`.
Test Plan: Used `transaction.search` to retrieve transactions of all new types.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18761
Summary: Ref T12689. See PHI178. This isn't a complete solution (you may still get mailed via packages/projects) but should fix the obvious issue, where "Resigned" reviewers are incorrectly always sent mail directly.
Test Plan: Had Alice resign, interacted as Bailey, no mail to Alice.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12689
Differential Revision: https://secure.phabricator.com/D18758
Summary:
Depends on D18756. Fixes T12539. See PHI190. Currently, when this occurs:
- Alice accepts.
- Bailey requests review.
- Alice views her dashboard.
...the revision appears in "Waiting on Other Reviewers" (regardless of whether other reviewers actually exist or not).
Instead, ignore these voided/non-current accepts and let the revisions appear in "Ready to Review", which is more natural.
Test Plan: Went through the steps above. On `master`, saw revision in "Waiting on Other Reviewers". After patch, saw it in "Ready to Review".
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12539
Differential Revision: https://secure.phabricator.com/D18757
Summary:
Ref T12539. See PHI190. Currently, each Diff has a `revisionID`, but Revisions do not point at the current active diff. To find the active diff for a given revision, we need to issue a separate query.
Furthermore, this query is inefficient for bulk loads: if we have a lot of revisions, we end up querying for all diff IDs for all those revisions first, then selecting the largest ones and querying again to get the actual diff objects. This strategy could likely be optimized but the query is a mess in any case.
In several cases, it's useful to have the active diff PHID without needing to do a second query -- sometimes for convenience, and sometimes for performance.
T12539 is an example of such a case: it would be nice to refine the bucketing logic (which only depends on active diff PHIDs), but it feels bad to make the page heavier to do it.
For now, this is unused. I'll start using it to fix the bucketing issue, and then we can expand it gradually to address other performance/convenience issues.
Test Plan:
- Ran migrations, inspected database, saw sensible values.
- Created a new revision, saw a sensible database value.
- Updated an existing revision, saw database update properly.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12539
Differential Revision: https://secure.phabricator.com/D18756
Summary: Give profile images a little more space, fix "/" spacing, add a tooltip.
Test Plan: {F5251205}
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18749
Summary:
Ref T2543. Fixes two relatively minor things:
- When builds finish in Harbormaster, send mail "From" the author.
- Set the `firstBroadcast` flag so that initial mail picks up earlier history (notably, the "reviewers" line).
For now, I'm not setting `firstBroadcast` on explicit "Request Review" (but maybe we should), and not trying to deal with weird cases where you leave a bunch of comments on a draft. Those might be fine as-is or may get tweaked later.
Test Plan: Created a revision with Harbormaster builds, ran builds, saw initial email come "From" the right user with more metadata.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18748
Summary: Depends on D18746. See PHI174. Adds small author portraits next to each blame line (this is similar to GitHub).
Test Plan:
My local test data isn't that great since I don't have commits from a lot of accounts, but looks functional:
{F5251056}
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18747
Summary:
Ref PHI174. This reverts most of these changes:
- 37843127e9 / D18481
- 94cad30ac3 / D18474
- 12ae08b6b1 / D18473
- 0a01334172 / D18462
- ac91ab1ef9 / D18452
These changes made the Diffusion blame view very similar to GitHub's blame view. See D18452 for a before/after of the bulk of these changes; the other revisions are bugfixes.
I think this was generally a step backward, and not motivated by solving a specific problem. I've found the new UI less usable than the old one, and at least one install (see PHI174) also has.
In particular, the revision/commit titles are very bulky and not terribly useful; the date column also isn't terribly useful; the "age" color actually IS pretty useful and was heavily de-emphasized.
I've kept one bugfix here (missing `'a'` tag type) and kept the upgraded icon for "Skip Past This Commit".
I'm going to follow this up with some additional changes:
- Show a small author profile icon, similar to GitHub, to address PHI174 more directly.
- Try a zebra-stripe on blocks of rows to make it more clear where changes affected by a particular commit begin and end.
- Try a hue shift, not just a brightness/saturation shift, to make the "age" color more distinct.
- Try computing colors as even steps, not based purely on age. Currently, if a file has one long-distant commit and several recent commits, all the recent ones show up as very bright green. I think this would probably be more useful if they were distributed more evenly across the available color bands.
Test Plan:
Viewed blame views in Diffusion, saw a more compact UI similar to the old UI.
{F5251019}
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18746
Summary: Ref T2543. After D18731, Herald build rules run more often, but now incorrectly try to run builds when Diffusion closes a revision because a commit landed.
Test Plan: Made some mundane updates locally; this is tricky to test comprehensively locally so I'm mostly planning to just push it to `secure`.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18745
Summary:
Ref T2543. Instead of autosubmitting revisions to "Needs Review" when builds finish, allow them to be held in "Draft" indefinitely.
There's currently no UI for this. I plan to just expose it as `arc diff --draft` for now, in a followup change.
Test Plan:
- Created a revision (via Conduit) with "hold as draft", saw it hold as draft after builds finished.
- Created a revision (normally), saw it autosubmit after builds finished.
- Requested review of a "hold as draft" revision to kick it out of draft state.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18737
Summary:
Depends on D18740. Prepares `arc` to receive a `--draft` flag by letting us switch to "differential.revision.edit" instead of "differential.createrevision".
To "differential.revision.edit", we need a transaction list, but we can't automatically construct this list from a field map. Return the transaction list alongside the field map.
The next change uses this list (if available) to switch us to the modern API method.
Test Plan: Ran `arc diff` on the experiemntal branch with followup changes, got a new revision.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18741
Summary: See PHI90. For now, this only provides a limited amount of information, but should satisfy the use case in PHI90 and build toward a more complete version in the future.
Test Plan: Used new Conduit method to retrieve information about diffs.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18744
Summary:
Ref T2543. Fixes T10109.
Currently, Herald only runs in Differential when a change updates the diff. This is partly for historical reasons, and partly because we don't want to restart builds every time someone makes a comment. However, this behavior is inconsistent with other applications (which always trigger on any change), and occasionally confusing to users (in T10109, for example) or otherwise undesirable.
A similar issue is that T2543 has introduced a "Draft" state, where revisions don't send normal mail until builds finish. This interacts poorly with "Send me an email" rules (which shouldn't do anything here) and particularly with "Send me an email + only run these actions the first time the rule matches", since that might have an effect like "do nothing when the revision is created, then never anything again since you already did nothing once".
To navigate both of these issues, let objects tell Herald that certain actions (like mail or builds) are currently forbidden. If a rule uses a field or action which is currently forbidden, the whole rule automatically fails before it executes, but doesn't count toward "only the first time" as far as Herald's tracking of rule execution is concerned.
Then, forbid mail for draft revisions, and forbid builds for revisions which didn't just get updated. Forbidding mail fixes the issues with "Send me an email" that were created by the introduction of the draft state.
Finally, make Herald run on every revision update, not just substantive updates to the diff. This resolves T10109.
Test Plan:
Created revisions via the draft -> submit workflow, saw different transcripts. Here's a mail action being forbidden for a draft:
{F5237324}
Here's a build action being forbidden for a "mundane" update:
{F5237326}
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T10109, T2543
Differential Revision: https://secure.phabricator.com/D18731
Summary: Depends on D18734. See PHI176. We run this query on the main Drydock lease web UI, among other places. There is currently no `status` key which can satisfy it.
Test Plan:
Viewed Drydock lease page to get the query.
Ran ##explain SELECT * FROM `drydock_lease` WHERE (status IN ('pending', 'acquired', 'active')) ORDER BY `id` DESC LIMIT 101;## before and after the change.
I don't have a ton of leases locally so the un-key'd EXPLAIN isn't //that// bad, but still shows that we're getting a better key. Before:
```
mysql> explain SELECT * FROM `drydock_lease` WHERE (status IN ('pending', 'acquired', 'active')) ORDER BY `id` DESC LIMIT 101;
+----+-------------+---------------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+---------------+-------+---------------+---------+---------+------+------+-------------+
| 1 | SIMPLE | drydock_lease | index | NULL | PRIMARY | 4 | NULL | 101 | Using where |
+----+-------------+---------------+-------+---------------+---------+---------+------+------+-------------+
1 row in set (0.00 sec)
```
After:
```
mysql> explain SELECT * FROM `drydock_lease` WHERE (status IN ('pending', 'acquired', 'active')) ORDER BY `id` DESC LIMIT 101;
+----+-------------+---------------+-------+---------------+------------+---------+------+------+---------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+---------------+-------+---------------+------------+---------+------+------+---------------------------------------+
| 1 | SIMPLE | drydock_lease | range | key_status | key_status | 130 | NULL | 5 | Using index condition; Using filesort |
+----+-------------+---------------+-------+---------------+------------+---------+------+------+---------------------------------------+
1 row in set (0.00 sec)
```
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Differential Revision: https://secure.phabricator.com/D18735
Summary: See PHI176. Depends on D18733. We issue a query when deleting files that currently doesn't hit any keys.
Test Plan:
Ran `./bin/remove destroy --force --trace F56376` to get the query.
Ran ##SELECT * FROM `file` WHERE storageEngine = 'blob' AND storageHandle = '23366' LIMIT 1## before and after the change.
Before:
```
mysql> explain SELECT * FROM `file` WHERE storageEngine = 'blob' AND storageHandle = '23366' LIMIT 1;
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
| 1 | SIMPLE | file | ALL | NULL | NULL | NULL | NULL | 33866 | Using where |
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
1 row in set (0.01 sec)
```
After:
```
mysql> explain SELECT * FROM `file` WHERE storageEngine = 'blob' AND storageHandle = '23366' LIMIT 1;
+----+-------------+-------+------+---------------+------------+---------+-------------+------+------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+------+---------------+------------+---------+-------------+------+------------------------------------+
| 1 | SIMPLE | file | ref | key_engine | key_engine | 388 | const,const | 190 | Using index condition; Using where |
+----+-------------+-------+------+---------------+------------+---------+-------------+------+------------------------------------+
1 row in set (0.00 sec)
```
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18734
Summary: See PHI176. This is issued periodically by the garbage collector. Normally this table is relatively small-ish so this missing key isn't hugely noticeable.
Test Plan:
Ran `./bin/garbage collect --collector daemon.processes --trace` to get the query the GC runs.
Ran ##DELETE FROM `daemon_logevent` WHERE epoch < 1508443504 LIMIT 100## before and after the key, saw a much better query plan afterward:
Before:
```
mysql> explain DELETE FROM `daemon_logevent` WHERE epoch < 1508443504 LIMIT 100;
+----+-------------+-----------------+------+---------------+------+---------+------+-------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-----------------+------+---------------+------+---------+------+-------+-------------+
| 1 | SIMPLE | daemon_logevent | ALL | NULL | NULL | NULL | NULL | 19325 | Using where |
+----+-------------+-----------------+------+---------------+------+---------+------+-------+-------------+
1 row in set (0.00 sec)
```
After:
```
mysql> explain DELETE FROM `daemon_logevent` WHERE epoch < 1508443504 LIMIT 100;
+----+-------------+-----------------+-------+---------------+-----------+---------+-------+------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-----------------+-------+---------------+-----------+---------+-------+------+-------------+
| 1 | SIMPLE | daemon_logevent | range | key_epoch | key_epoch | 4 | const | 1 | Using where |
+----+-------------+-----------------+-------+---------------+-----------+---------+-------+------+-------------+
1 row in set (0.00 sec)
```
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18733
Summary: See PHI176. We issue a query with only `artifactIndex` from `BuildTarget`, but don't have an applicable key.
Test Plan: This isn't on the normal Harbormaster execution path so I'm not 100% sure I have a local repro, but will confirm with customer.
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Differential Revision: https://secure.phabricator.com/D18732
Summary: See PHI177. Ref T12974. PonderQuestion was overlooked during the Ferret engine conversions.
Test Plan:
Ran migrations, searched for questions, got results:
{F5241185}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12974
Differential Revision: https://secure.phabricator.com/D18736
Summary:
Ref T2543. When Harbormaster finishes builds and promotes a draft revision to review, we currently publish "Harbormaster requested review of...".
Instead, attribute this action to the author, since that's more natural and more useful.
Test Plan: Promoted a diff locally, saw it attributed to me rather than Harbormaster.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18730
Summary:
See PHI165. See D18715. These objects (projects, blogs) also need implementations now.
(I thought about making this method `abstract` or doing try/catch to maybe make this more robust, but I think this should be the end of it, and those changes have mild complexity/compatibility/risk issues.)
Test Plan: Changed `bin/search index` to index only one document of each type, ran `bin/search index --all --force`, saw no more comment-related errors.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18729
Summary:
Ref T12974. Ferret object queries SELECT a virtual "_ft_rank" column for relevance ordering.
Currently, they always SELECT this column. That's fine and doesn't hurt anything, but makes developing and debugging things kind of a pain since every query has this `, blah blah _ft_rank` junk.
Instead, construct this column only if we're actually going to use it.
Mostly, this cleans up DarkConsole / query logs a bit.
Test Plan:
Viewed normal query results on various pages, viewed global search results, ordered Maniphest tasks by normal stuff and by "Relevance".
Viewed DarkConsole, saw no more "_ft_rank" junk on normal pages.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12974
Differential Revision: https://secure.phabricator.com/D18728
Summary: Ref T12680. Use the slightly sleeker construction from D18722 in Differential.
Test Plan: Viewed revision list, reordered by date modified.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12680
Differential Revision: https://secure.phabricator.com/D18727
Summary:
Ref T12680. See PHI167. See that task for discussion.
Rewrite `DiffusionCommitQuery` to work more like `DifferentialRevisionQuery`, and use a UNION to find "all revisions you need to audit OR respond to".
I tried to get this working a little more cleanly than RevisionQuery does, and can probably simplify that now.
Test Plan: Poked at the UI locally without hitting any apparent issues, but my local data is pretty garbage at this point. I'll take a look at how the query plans work on `secure`.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12680
Differential Revision: https://secure.phabricator.com/D18722
Summary:
Ref T2543. When a revision is created, we check if any builds are waiting/failed, and submit it for review immediately if we aren't waiting for anything.
In doing this, we ignore builds with only autotargets, since these are client-side and failures from local `arc lint` / `arc unit` should not count (the user has already chosen to ignore/skip them).
The way we do this has some issues:
- Herald may have started builds, but they may still be PENDING and not have any targets yet. In this case, we'll see "no non-autotargets" and ignore the build, which is wrong.
- We have to load targets but don't really care about them, which is more work than we really need to do.
- And it's kind of complex, too.
Instead, just let `BuildQuery` filter out "autobuilds" (builds generated from autoplans) with a JOIN.
Test Plan: Ran `arc diff` with builds configured, got a clean "Draft" state instead of an incorrect promotion directly to "Needs Review".
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18721
Summary:
See PHI158. In the RefEngine, we test if any old branch positions have been removed from the repository. This is uncommon (but not impossible) in Mercurial, and corresponds to users deleting branches in Git.
Currently, we end up running `hg log` for each position, in parallel. Because of Python's large startup overhead, this can be resource intensive for repositories with a large number of branches.
We have to do this in the general case because the caller may be asking us to resolve `tip`, `newfeature`, `tip~3`, `9`, etc. However, in the specific case where the refs are 40-digit hashes, we can bulk resolve them if they exist, like this:
```
hg log ... --rev (abcd or def0 or ab12 or ...)
```
In the general case, we could probably do less of this than we currently do (instead of testing all old heads, we could prune the list by removing commits which we know are still pointed to by current heads) but that's a slightly more involved change and the effect here is already dramatic.
Test Plan:
Verified that CPU usage drops from ~110s -> ~0.9s:
Before:
```
epriestley@orbital ~/dev/phabricator $ time ./bin/repository refs nss
Updating refs in "nss"...
Done.
real 0m14.676s
user 1m24.714s
sys 0m21.645s
```
After:
```
epriestley@orbital ~/dev/phabricator $ time ./bin/repository refs nss
Updating refs in "nss"...
Done.
real 0m0.861s
user 0m0.882s
sys 0m0.213s
```
- Manually resolved `blue`, `tip`, `9`, etc., got expected results.
- Tried to resolve invalid hashes, got expected result (no resolution).
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18717
Summary: Most of this document is no longer relevant, since we're happy to work on prototypes if you're paying us and no longer have any meaningful free support.
Test Plan: Read document.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18719
Summary: See rPcd14194a329788d5fff6365bcade278fd18f3612 for a similar change. Implement `getApplicationTransactionCommentObject()` to return `null` explicitly.
Test Plan: Ran `bin/search index --type ownerspackage`, got indexing after change.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18715
Summary: Ref T2543. When revisions are in the draft state, tell the user what we're waiting for or why they aren't moving forward.
Test Plan: {F5228840}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18714
Summary: Ref T2543. This is a less ambitious version of the rule in D18628, which I backed off from, since I think this probably still has a fair number of loose ends to tie up.
Test Plan: Created a revision locally.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18713
Summary:
Ref T13000. Since other changes have generally made the ngrams table manageable, I'm not planning to enable common ngrams by default at this time.
Instead, make the threshold configurable with "--threshold" so we can guide installs through tuning this if they want (e.g. PHI110), and tune hosted instances.
(This might eventually become automatic, but just smoothing this bit off for now feels reasonable to me.)
Test Plan: Ran with `--reset`, and with various invalid and valid `--threshold` arguments.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18710
Summary:
Ref T13000. The new approach for dumping database-by-database means that we don't get CREATE DATABASE or USE statements, which makes importing the dump again inconvenient.
Manually stitch these into the dump.
Test Plan:
- Used `bin/storage dump --namespace ...` to dump a smaller local instance.
- Used `bin/storage destroy --namespace ...`, to destroy the namespace, then inported the dump cleanly.
- Verified that each CREATE DATABASE statement appears only once.
- Verified that `bin/storage renamespace --live` can correctly process this file.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18707
Summary: See PHI147.
Test Plan: Called the method from the web UI, got sensible results.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18706
Summary:
Depends on D18702. Ref T13008. This replaces the old hard-coded single rate limit with multiple flexible limits, and defines two types of limits:
- Rate: reject requests if a client has completed too many requests recently.
- Connection: reject requests if a client has too many more connections than disconnections recently.
The connection limit adds +1 to the score for each connection, then adds -1 for each disconnection. So the overall number is how many open connections they have, at least approximately.
Supporting multiple limits will let us do limiting by Hostname and by remote address (e.g., a specific IP can't exceed a low limit, and all requests to a hostname can't exceed a higher limit).
Configuring the new limits looks something like this:
```
PhabricatorStartup::addRateLimit(new PhabricatorClientRateLimit())
->setLimitKey('rate')
->setClientKey($_SERVER['REMOTE_ADDR'])
->setLimit(5);
PhabricatorStartup::addRateLimit(new PhabricatorClientConnectionLimit())
->setLimitKey('conn')
->setClientKey($_SERVER['REMOTE_ADDR'])
->setLimit(2);
```
Test Plan:
- Configured limits as above.
- Made a lot of requests, got cut off by the rate limit.
- Used `curl --limit-rate -F 'data=@the_letter_m.txt' ...` to upload files really slowly. Got cut off by the connection limit. With `enable_post_data_reading` off, this correctly killed the connections //before// the uploads finished.
- I'll send this stuff to `secure` before production to give it more of a chance.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13008
Differential Revision: https://secure.phabricator.com/D18703
Summary:
Ref T13008. Depends on D18701. The overall goal here is to make turning `enable_post_data_reading` off not break things, so we can run rate limiting checks before we read file uploads.
The biggest blocker for this is that turning it off stops `$_FILES` from coming into existence.
This //appears// to mostly work. Specifically:
- Skip the `max_post_size` check when POST is off, since it's meaningless.
- Don't read or scrub $_POST at startup when POST is off.
- When we rebuild REQUEST and POST before processing requests, do multipart parsing if we need to and rebuild FILES.
- Skip the `is_uploaded_file()` check if we built FILES ourselves.
This probably breaks a couple of small things, like maybe `__profile__` and other DarkConsole triggers over POST, and probably some other weird stuff. The parsers may also need more work than they've received so far.
I also need to verify that this actually works (i.e., lets us run code without reading the request body) but I'll include that in the change where I update the actual rate limiting.
Test Plan:
- Disabled `enable_post_data_reading`.
- Uploaded a file with a vanilla upload form (project profile image).
- Uploaded a file with drag and drop.
- Used DarkConsole.
- Submitted comments.
- Created a task.
- Browsed around.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13008
Differential Revision: https://secure.phabricator.com/D18702
Summary:
Ref T13000. Garbage collecting common ngrams is slow because MySQL isn't all that great at deleting rows quickly. See PHI96, where it looks like it's going to take a week to GC ngrams for a ~million objects at a relatively conservative 0.15 threshold.
In the event of a restore, we can reduce the impact by persisting this table so the ngrams just don't get built when the reindex happens.
Test Plan: Viewed schema in Config, saw common ngrams tables marked as "Data" instead of "Index".
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18696
Summary: Noticed a couple of typos in the docs, and then things got out of hand.
Test Plan:
- Stared at the words until my eyes watered and the letters began to swim on the screen.
- Consulted a dictionary.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, yelirekim, PHID-OPKG-gm6ozazyms6q6i22gyam
Differential Revision: https://secure.phabricator.com/D18693
Summary:
Ref PHI109. Ref T11786. We currently test elapsed time every 64 iterations (since iterations are normally very fast), but at least one install is seeing the page timeout after 30 seconds.
One reason could be that cache fills may occur, and are likely to be much slower than normal iterations. In an extreme case, we could do 64 cache fills before checking the time. Tweak thing so that we always check the time after doing a cache fill, regardless of how many iterations have elapsed since the last attempt.
Additionally, this API method currently accepts an arbitrary number of paths, but implicitly limits each cache query to 500ms. If more than 60 paths are passed, this may exceed 30s. Only let the cache churn for a maximum of 10s across all paths.
If this is more the latter issue than the former, this might replace the GraphCache timeouts with `git` timeouts, but at least our understanding of what's going on here will improve.
Test Plan: This is difficult to test convincingly locally, since I can't reproduce the original issue. It still works after these changes, but it worked fine before these changes too.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T11786
Differential Revision: https://secure.phabricator.com/D18692
Summary:
Ref T13000. Currently, queries can only be executed from the web UI, which requires logging in as a user. I really want to avoid doing that wherever we can, but being able to execute queries on an instance (and, particularly, see the ngrams and timings on the underlying lookups) would have been helpful in several cases.
Improve tooling a bit in advance of the "common ngrams" stuff going out since it seems likely that it will be useful if issues arise.
Test Plan: Ran `bin/search query --query ...`, got useful minimal output. Ran with `--trace` to get internals.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18690
Summary:
Ref T13000. After an ngram is marked as "common", we can delete it from the storage table.
Currently, the only way to get ngrams marked as "common" is to manually run `bin/search ngrams`, so this has no impact on normal installs.
Test Plan: Ran `bin/garbage collect`, saw it start chewing through my local Maniphest ngrams table and removing common ngrams.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18687
Summary:
Ref T13000. This marks each table as either "data" (normal data), "cache" (automatically rebuilt, no need to ever dump) or "index" (can be manually rebuilt).
By default, `bin/storage dump` dumps data and index tables, but not cache tables.
With `--no-indexes`, it dumps only data tables. Indexes can be rebuilt after a restore with `bin/search index --all ...`.
Test Plan:
- Ran `--no-indexes` and normal dumps with `--trace`, verified that cache and index (former case) or cache only (latter case) tables were dumped with `--no-data`.
- Verified dump has the same number of `CREATE TABLE` statements as before the changes.
- Reviewed persistence tags in the web UI (note Ferret engine tables are "Index"):
{F5210886}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18682
Summary:
Ref T13000. This allows us to be more selective about which tables we dump data for, to reduce the size of backups and exports. The immediate goal is to make large `ngrams` tables more manageable in the cluster, but this generally makes all backups and exports faster and easier.
Here, tables are dumped one at a time. A followup change will sometimes add the `--no-data` flag, to skip dumping readthrough caches and (optionally) rebuildable indexes.
Test Plan: Compared a dump from `master` and from this branch, found them to be essentially identical. The new dump has a little more header information in each section. Verified each contains the same number of `CREATE TABLE` statements.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18679
Summary: See PHI118. Enables hovercards to support peeking at tags and other details if you, e.g., create numerous identical subtasks of each task.
Test Plan: {F5210816}
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18681
Summary:
If `account.editable` is set to false, we try to add a `null` button and fatal:
> Argument 1 passed to PHUIHeaderView::addActionLink() must be an instance of PHUIButtonView, null given, called in /srv/phabricator/phabricator/src/applications/settings/panel/PhabricatorSettingsPanel.php on line 290
Instead, don't try to render `null` as a button.
Test Plan:
- Configured `account.editable` false.
- Viewed email address settings.
- Before: fatal.
- After: page works, no button is provided.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18677
Summary:
Depends on D18672. Ref T13000. This does an on-demand build of the common ngrams table.
Plan here is:
- Push to `secure`.
- Build the common ngrams table here.
- See if stuff breaks?
If it looks okay on this dataset, we can build out the GC support and try it in production.
Test Plan:
- Locally, my dataset has a bunch of `bin/lipsum` tasks with similar, common words.
- Verified that ipsum terms now skip ngrams. For "lorem ipsum" search performance actually IMPROVED by skipping the ngrams table (12s to 9s).
- Queried for normal terms, got very fast results using the ngram table, as normal.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18673
Summary:
Ref T13000. This adds support for tracking "common" ngrams, which occur in too many documents to be useful as part of the ngram index.
If an ngram is listed in the "common" table, it won't be written when indexing documents, or queried for when searching for them.
In this change, nothing actually writes to the "common" table. I'll start writing to the table in a followup change.
Specifically, I plan to do this:
- A new GC process updates the "common" table periodically, by writing ngrams which appear in more than X% of documents to it, for some value of X, if there are at least a minimum number of documents (maybe like 4,000).
- A new GC process deletes ngrams that have been added to the common table from the existing indexes.
Hopefully, this will pare down the ngrams index to something reasonable over time without requiring any manual tuning.
Test Plan:
- Ran some queries and indexes.
- Manually inserted ngrams `xxx` and `yyy` into the ngrams table, searched and indexed, saw them ignored as viable ngrams for search/index.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T13000
Differential Revision: https://secure.phabricator.com/D18672
Summary:
See PHI112. The install presumably wants to generate links to Diffusion commits from an external tool, but only knows the short name of the repository.
Provide a `/source/phabricator/commit/abcdef908273` URI which redirects to the canonical URI for the commit.
Test Plan:
- Visited `/source/` URI for a commit, got a redirect.
- Visited normal URI for a commit, got a commit page.
- Visited `/branches/` and `/tags/` for a `/source/` repository, got proper pages.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18676
Summary:
See PHI94. I considered this initially but wasn't sure about it. However, PHI94 brings up the good point that we already use a similar rule in Maniphest.
For consistency, only show visible columns here too.
Test Plan: Used "Move tasks to column..." on a board with visible and hidden columns, only saw visbile columns offered in the dropdown.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18668
Summary:
See <https://discourse.phabricator-community.org/t/daemons-tasks-crashing-in-a-loop-during-reindex/506/1>. Some object types (for example, Passphrase Credentials) support indexing but not commenting.
Make `withComments(...)` work properly if the transaction type does not support comments.
Test Plan:
Indexed a credential (no comments) and a revision (comments) with `bin/search index --trace ...`.
Before, credential fataled.
After, credetial succeeds, and skips the transaction query.
Before and after, the revision queries the transaction table.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18667
Summary:
Ref T5523. See PHI50. See PHI40. This isn't perfect, but should improve things.
Add a "Move tasks to column..." action to workboards which moves all visible tasks in a column to another column, either on the same board or on a different board.
This is a two-step process so that I don't have to write Javascript, and because I'm not 100% sure this is what users actually want/need. If it sticks, the UI could be refined later.
- The first dialog asks you to choose a project, defaulting ot the current project.
- The second dialog asks you to choose a column on that project's board.
Test Plan:
- Moved tasks on the same board.
- Moved tasks to a different board.
- Tried to move tasks to the starting column, got a sensible error.
- Tried to move tasks to no project, got a sensible error.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T5523
Differential Revision: https://secure.phabricator.com/D18665
Summary:
Ref PHI91. When Owners (or Herald, or manual user action) adds package reviewers to a revision, later updates to the revision make some of them less relevant or irrelevant.
Provide a hint when a package reviewer doesn't own any of the paths that a diff changes. Humans can then decide if the reviewer is obsolete/irrelevant or not.
This is a rough cut to get the feature working, design could probably use some tweaking if it sticks.
Test Plan: {F5204309}
Reviewers: amckinley
Reviewed By: amckinley
Subscribers: jboning
Differential Revision: https://secure.phabricator.com/D18663
Summary:
Ref PHI101. It looks like this was maybe copy/pasted by mistake in recent design refactoring.
We need to pass the full path, not the `basename()` of the path, to the search form.
Test Plan: Searched inside `scripts/test/`, found results inside `scripts/test/`.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18664
Summary:
Ref T12997. Although we can't query by transaction type (since we can't easily enumerate all possible types which may have comments -- inline types may also have comments), we //can// just check if there's a comment row or not.
This reduces the amount of garbage we need to load to rebuild indexes for unusual objects with hundreds and hundreds of mentions.
Test Plan:
- Used batch editor to mention a task 700 times.
- Indexed it before and after this change, saw index time drop from {nav 1600ms > 160ms}.
- Made some new comments on it, verified that they still indexed/queried properly.
- Browsed around, made normal transactions, made inline comments.
- Added a unique word to an inline comment, indexed revision, searched for word, found revision.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12997
Differential Revision: https://secure.phabricator.com/D18660
Summary:
Ref T12997. See that task for more details. Briefly, an unusual dataset (where commits are mentioned hundreds of times by other commits) is causing some weird memory behavior in the daemons.
Forcing PHP to GC cycles explicitly after each task completes seems to help with this, by cleaning up some of the memory between tasks. A more thorough fix might be to untangle the `$xactions` structure, but that's significantly more involved.
Test Plan:
- Did this locally in a controlled environment, saw an immediate collection of a 500MB `$xactions` cycle.
- Put a similar change in production, memory usage seemed less to improve. It's hard to tell for sure that this does anything, but it shouldn't hurt.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12997
Differential Revision: https://secure.phabricator.com/D18657
Summary:
See PHI87. Ref T12974. The `array_slice()` method of splitting the string apart can perform poorly for large input strings. I think this is mostly just the large number of calls plus building and returning an array being not entirely trivial.
We can just use `substr()` instead, as long as we're a little bit careful about keeping track of where we're slicing the string if it has UTF8 characters.
Test Plan:
- Created a task with a single, unbroken blob of base64 encoded data as the description, roughly 100KB long.
- Saw indexing performance improve from ~6s to ~1.5s after patch.
- Before: https://secure.phabricator.com/xhprof/profile/PHID-FILE-nrxs4lwdvupbve5lhl6u/
- After: https://secure.phabricator.com/xhprof/profile/PHID-FILE-6vs2akgjj5nbqt7yo7ul/
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12974
Differential Revision: https://secure.phabricator.com/D18649
Summary:
See PHI87. Ref T12974. Currently, we do a lot more work here than we need to: we call `phutil_utf8_strtolower()` on each token, but can do it once at the beginning on the whole block.
Additionally, since ngrams don't care about order, we only need to convert unique tokens into ngrams. This saves us some `phutil_utf8v()`. These calls can be slow for large inputs.
Test Plan:
- Created a ~4MB task description.
- Ran `bin/search index Txxx --profile ...` to profile indexing performance before and after the change.
- Saw total runtime drop form 38s to 9s.
- Before: <https://secure.phabricator.com/xhprof/profile/PHID-FILE-wiht5d7lkyazaywwxovw/>
- After: <https://secure.phabricator.com/xhprof/profile/PHID-FILE-efxv56q2hulr6kjrxbx6/>
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12974
Differential Revision: https://secure.phabricator.com/D18647
Summary:
Fixes T12995. Currently, the result highlighter (which shows //where// terms matched) only works in "term" mode, not in "substring" mode.
Provide better feedback and behvaior:
- When a term is a substring term, color it a little differently and add a tooltip. (This is partly to make it easier to debug/diagnose things, probably not enormously valuable to users.)
- When a term is a substring term, highlight it anywhere in the results.
Test Plan:
Queried for latin and CJK terms.
Here is CJK being highlighted:
{F5192195}
Here is substring vs non-substring implicit behavior:
{F5192196}
Here's ONLY terms being highlighted:
{F5192198}
Here's terms and substrings, since the query now has a substring:
{F5192201}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12995
Differential Revision: https://secure.phabricator.com/D18635
Summary:
Ref T2543. This doesn't stand alone since mail still goes out normally, but gets this piece working: new revisions start as "Draft", then after updates if there are no builds they go into "Needs Review".
This should work in general because builds update revisions when they complete, to publish a "Harbormaster finished build yada yada" transaction. So either we'll un-draft immediately, or un-draft after the last build finishes.
I'll hold this until the mail and some other stuff (like UI hints) are in slightly better shape since I think it's probably too rough on its own.
Test Plan: Created revisions locally, saw them un-draft after builds.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18628
Summary:
Ref T2543. Currently, we always do some special things when a revision is created, mostly adding more stuff to the mail.
With drafts, we want to suppress initial mail and send this big, rich mail only when the revision actually moves out of "draft".
Prepare the code for this, with the actual methods hard-coded to the current behavior. This will probably take some tweaking but I think I got most of it.
Test Plan: Banged around in Differential so it sent some mail, saw normal mail without anything new.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18627
Summary:
Ref T2543. Most actions are not available for drafts.
Authors can "Request Review" (move out of draft to become a normal revision) or "Abandon".
Non-authors can't do anything (maybe we'll let them do something later -- like "Commandeer"? -- if there's a good reason).
Test Plan: Viewed a draft revision as an author and non-author, saw fewer actions available.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18626
Summary: See PHI79. When you edit another user's SSH keys (normally, for a bot account) we currently redirect you to an older URI.
Test Plan:
- Viewed a bot's profile page.
- Clicked "Edit Settings" on the Manage page.
- Went to "SSH Keys".
- Uploaded an SSH key.
- Before: redirected to a 404 after finishing the workflow.
- After: redirected to the same page after the workflow.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18633
Summary:
See PHI78. The user was getting this message and (reasonably) interpreted it to mean "reset mail can never be sent to unverified addresses".
Reword it to be more clear, albeit an entire paragraph long. I don't really have a good solution in these cases where we'd need a whole page to explain what's happening (this, plus "we can't tell you which address you should use because an attacker could get information if we did" and "this rule defuses the risk that an opportunistic attacker may try to compromise your account after you add an email you don't own by mistake"). We could write it up separately and link to it, but I feel like that stuff tends to get out of date.
Just land somewhere in the middle.
Test Plan: {F5189105}
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18630
Summary:
Ref T2543. There's no way to put revisions into this state yet, but start adding support for when there is.
Adds the status constant, plus support for bucketing them.
Test Plan:
- Manually put a revision in "Draft" state by updating the database directly.
- Verified my drafts showed up in a "Drafts" section on the bucket view.
- Verified others' drafts did not appear on the action bucket view.
- Viewed revisions, queried for "Draft" revisions, etc (stuff we get for free).
{F5186781}
{F5186782}
{F5186783}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T2543
Differential Revision: https://secure.phabricator.com/D18625
Summary:
Fixes T12986. I caught this bug in the changes from D18584: when we moved a large hunk to file storage, we would decompress it but keep the "deflated" flag. This could cause confusion when loading it later. I missed this in testing since I wasn't exhaustive enough in checking hunks and didn't run into a compressed one.
Instead of compressing on `save()`, compress during the normal workflow.
We currently never advise users to run this workflow so I didn't bother trying to clean up possible existing migrations.
Test Plan:
- Ran `bin/differential migrate-hunk` on compressed hunks, moving them to and from file storage. Saw them work correctly and remain compressed.
- Created new small (uncompressed) and large (compressed) hunks, verified they work properly and get compressed (if applicable).
- Used `bin/cache purge --caches changeset` to clear changeset caches and make sure the actual table was being hit.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T12986
Differential Revision: https://secure.phabricator.com/D18624
Summary: Despite how I (and everyone else?) pronounce it, it is spelled with an "a". See PHI38.
Test Plan: Googled both spellings.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18622
Summary:
See PHI71. These didn't get properly updated when we wrote Subprojects and Milestones, and should use materialized members, not raw members. Swap the query so projects you are an indirect member of (e.g., milestones you are a member of the parent for, and parent projects you are a member of a subproject of) are included in the result list.
Also fix a bad typeahead datasource.
Test Plan:
- Ran a dry run with the test console, saw project PHIDs for milestones and parent projects in the raw field value.
- Tried to set "Author's projects" to a user, no longer could.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18619
Summary:
Ref T11823. I think this is the last callsite which relies on the old data format: `bin/repository parents` rebuilds a cache which we don't currently use very heavily.
Update it to work with the new data.
Test Plan: Ran `bin/repository parents <repository> --trace`, saw successful script execution and reasonable-looking output.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T11823
Differential Revision: https://secure.phabricator.com/D18615
Summary:
Ref T11823. This is the meaty part of the change, and updates `RefEngine` to use separate RefCursor (for names) and RefPosition (for actual commit positions) tables.
I'll hold this whole series until after the release cut so it has some time to bake on `secure` to look for issues. It's also not a huge problem if there are bugs here since these tables are just caches anyway, although they do feed into some other things, and obviously it's never good to have bugs.
Test Plan:
- This logic can be invoked directly with `bin/repository refs <repository> --trace --verbose`.
- Ran that on unchanged repositories, new branches, removed branches, and modified branches. Saw appropriate output and cursor positions.
- Ran on a mercurial repository to test the close/open logic, saw it correct open/closed state of incorrect positions.
- Browed around Diffusion in various repositories.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T11823
Differential Revision: https://secure.phabricator.com/D18614
Summary:
Ref T11823. This change isn't standalone, but prepares for the more involved code change by dropping obsolete columns from the RefCursor table and adding the unique key we need to prevent the ambiguous/duplicate refs issue.
This data was moved to the RefPosition table in D18612.
Test Plan: Ran storage upgrade. See next revision for more substantial testing of this change series.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T11823
Differential Revision: https://secure.phabricator.com/D18613
Summary:
Ref T11823. Currently, we have a "RefCursor" table which stores rows like `<branch or tag name, commit it is pointing at>` with some more data.
Because Mercurial can have a single branch pointing at several different places, this table must allow multiple rows with the same branch or tag name.
Among other things, this means there isn't a single PHID which can be used to identify a branch name in a stable way. However, we have several UIs where we want to be able to do this.
Some specific examples where we run into trouble: in Mercurial, if there are 5 heads for "default", that means there are 5 phids. And currently, if someone deletes a branch, we lose the PHID for it. Instead, we'd rather retain it so the whole world doesn't break if you accidentally delete a branch and then fix it a little later.
(I'll likely hold this until the rest of the logic is fleshed out a little more in followup changes.)
Test Plan: Ran `bin/storage upgrade`, saw the table get created without warnings.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T11823
Differential Revision: https://secure.phabricator.com/D18602
Summary:
See <https://discourse.phabricator-community.org/t/unable-to-use-current-mercurial-on-debian-stretch/391>.
The Mercurial commit is helpful in particular: <https://www.mercurial-scm.org/repo/hg/rev/77eaf9539499>
We weren't vulnerable to the security issue (users can not control any part of the command) but pass the working directory explicitly to get past the new safety check.
I left `setCWD()` in place (a few lines below) just because it can't hurt, and in some other contexts it sometimes matter (for example, if commit hooks execute, they might inherit the parent CWD here or in other VCSes).
Test Plan:
- Cloned from a Mercurial repo locally over HTTP.
- Verified that SSH cloning already uses `-R` (it does, see `DiffusionMercurialServeSSHWorkflow`).
- Did not actually upgrade to Mercurial 4.0/4.1.3 to completely verify this, but a user in the Discourse thread asserted that a substantially similar fix worked correctly.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18611
Summary: See T12414. This just gets started; we still need edit endpoints for network interfaces and bindings.
Test Plan: Created some devices/services from the conduit UI.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D18605
Summary:
See PHI66. See that issue for context. This UI is bad broken legacy junk, but was especially broken when reporting merges.
These do not currently generate a "status" transaction, so they were never counted as task closures. Pretend they're normal closures.
This is still wrong, but should be much closer to the real numbers. Specifically, if you merge a closed task into another task, it will incorrectly be counted as an extra close. This could result in negative tasks, but the numbers should be much closer to reality than they are today even so.
The "Facts" application (T1562) is the real pathway forward here in the longer term.
Test Plan:
- Moved my `maniphest_transactions` table aside with `RENAME TABLE ...`.
- Created a new empty table with `CREATE TABLE ... LIKE ...`.
- Reloaded reports UI, saw empty chart.
- Created, closed, and reopened tasks while reloading the chart, saw accurate reporting.
- Merged an open task into another task, saw bad reporting.
- Applied patch, saw the right chart again.
Reviewers: amckinley
Reviewed By: amckinley
Differential Revision: https://secure.phabricator.com/D18601
Summary:
Ref T11823. See PHI68. T11823 has a full description of this issue and a plan to fix it, but the full plan is relatively complicated.
Until that can happen, provide a workaround for the biggest immediate issue, where multiple copies of a ref cursor can cause `executeOne()` to throw, since it expects a single result. In practice, these copies are always identical so we can just pick the first one.
This will get cleaned up once T11823 is fixed properly.
Test Plan:
Forced the table into a duplicate/ambiguous state, reproduced a similar-looking error:
{F5180999}
Applied the patch, got the "Land" to work as expected:
{F5181000}
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T11823
Differential Revision: https://secure.phabricator.com/D18599
Summary:
Ref T12033. This is a very narrow fix for this issue, but it should fix the major error: don't attach patches if they're bigger than the mail body limit (by default, 512KB).
Specifically, the logs from an install in T12033 show a 112MB patch being attached, and that's the biggest practical problem here.
I'll follow up on the tasks with more nuanced future work.
Test Plan: Enabled `differential.attach-patches`, saw a patch attached to email. Set the byte limit very low, saw patches get thrown away.
Reviewers: chad, amckinley
Reviewed By: amckinley
Maniphest Tasks: T12033
Differential Revision: https://secure.phabricator.com/D18598
