Summary:
Fixes T13468. See that task for discussion. The older source-rendering code mixes "line number" / "1-based" lists with "block number" / "0-based" lists and then has other bugs which cancel this out.
For block-based diffs, build an explicit block-based mask with only block numbers. This sort of sidesteps the whole issue.
Test Plan: Viewed the diff with the original reproduction case, plus various other block-based diffs, including one-block image diffs, in unified and side-by-side mode. Didn't spot any oddities.
Maniphest Tasks: T13468
Differential Revision: https://secure.phabricator.com/D20959
Summary:
Fixes T13475. Sometimes, we issue a "no op" / "default permit" / "unchallenged" MFA token, when a user with no MFA configured does something which is configured to attempt (but not strictly require) MFA.
An example of this kind of action is changing a username: usernames may be changed even if MFA is not set up.
(Some other operations, notably "Sign With MFA", strictly require that MFA actually be set up.)
When a user with no MFA configured takes a "try MFA" action, we see that they have no factors configured and issue a token so they can continue. This is correct. However, this token causes the assocaited timeline story to get an MFA badge.
This badge is incorrect or at least wildly misleading, since the technical assertion it currently makes ("the user answered any configured MFA challenge to do this, if one exists") isn't explained properly and isn't useful anyway.
Instead, only badge the story if the user actually has MFA and actually responded to some kind of MFA challege. The badge now asserts "this user responded to an MFA challenge", which is expected/desired.
Test Plan:
- As a user with no MFA, renamed a user. Before patch: badged story. After patch: no badge.
- As a user with MFA, renamed a user. Got badged stories in both cases.
Maniphest Tasks: T13475
Differential Revision: https://secure.phabricator.com/D20958
Summary: Fixes T13480. Adds the remaining missing Owners package rules for Herald commit adapters.
Test Plan: Created hooks which care about these fields, pushed commits, saw sensible transcript values.
Maniphest Tasks: T13480
Differential Revision: https://secure.phabricator.com/D20957
Summary: Ref T13480. The Herald "Commit" rules still use raw commit data properties to identify authors and committers. Instead, use repository identities.
Test Plan: Wrote a Herald rule using all four fields, ran it against various commits with and without known authors. Checked transcript for sensible field values.
Maniphest Tasks: T13480
Differential Revision: https://secure.phabricator.com/D20955
Summary:
Ref T13480. Currently, Herald commit hook rules use a raw address resolution query to identify the author and committer for a commit. This will get the wrong answer when the raw identity string has been explicitly bound to some non-default user (most often, it will fail to identify an author when one exists).
Instead, use the "IdentityEngine" to properly resolve identities.
Test Plan: Authored a commit as `X <y@example.com>`, a raw identity with no "natural" matches to users (e.g., no user with that email or username). Bound the identity to a particular user in Diffusion. Wrote a Herald pre-commit content rule, pushed the commit. Saw Herald recognize the correct user when evaluating rules.
Maniphest Tasks: T13480
Differential Revision: https://secure.phabricator.com/D20953
Summary:
Ref T13480. Currently, some Herald field types are rendered in an unfriendly way on transcripts. Particularly, PHID lists are rendered as raw PHIDs.
Improve this by delegating rendering to Value objects and letting "PHID List" value objects render more sensible handle lists. Also improve "bool" fields a bit and make more fields render an explicit "None" / empty value rather than just rendering nothing.
Test Plan: Viewed various transcripts, including transcripts covering boolean values, the "Always" condition, large blocks of text, and PHID lists.
Maniphest Tasks: T13480
Differential Revision: https://secure.phabricator.com/D20951
Summary:
Ref T13480. Currently, when Herald renders a transcript, it puts display labels into array keys. This is a bad pattern for several reasons, notably that the values must be scalar (so you can't add icons or other markup later) and the values must be unique (which is easily violated because many values are translated).
Instead, keep values as list items.
Test Plan: Viewed Herald transcripts, saw no (meaningful) change in rendering output.
Maniphest Tasks: T13480
Differential Revision: https://secure.phabricator.com/D20949
Summary: Ref T13480. When Herald renders rules, it partly uses a very old handle pre-loading mechanism where PHIDs are extracted and loaded upfront. This was obsoleted a long time ago and was pretty shaky even when it worked. Get rid of it to simplify the code a little.
Test Plan: Viewed Herald rules rendered into static text with PHID list actions, saw handles. Grepped for all affected methods.
Maniphest Tasks: T13480
Differential Revision: https://secure.phabricator.com/D20948
Summary: Fixes T13482. Although this style makes physical sense by relationship to a written checklist, it seems to do more harm than good in practice.
Test Plan: Wrote a checklist with a checked-off item in remarkup, saw no more line-through.
Maniphest Tasks: T13482
Differential Revision: https://secure.phabricator.com/D20954
Summary: Ref T13480. Add an "Author's packages" field to Herald to support writing rules like "if affected packages include X, and author's packages do not include X, raise the alarm".
Test Plan: Wrote and executed rules with the field, saw a sensible field value in the transcript.
Maniphest Tasks: T13480
Differential Revision: https://secure.phabricator.com/D20947
Summary: Ref T13480. These fields don't serve a specific strong use case, but are broadly reasonable capabilities after "state" vs "change" actions were relaxed by T13283.
Test Plan: Wrote rules using the new fields, added and removed projects (and did neither) to fire them / not fire them. Inspected the transcripts to see the project PHIDs making it to the field values.
Maniphest Tasks: T13480
Differential Revision: https://secure.phabricator.com/D20946
Summary:
Fixes T13479. The behavior of "git rev-parse --show-toplevel" has changed in Git 2.25.0, and it now fails in bare repositories.
Instead, use "git rev-parse --git-dir" to sanity-check the working copy. This appears to have more stable behavior across Git versions, although it's a little more complicated for our purposes.
Test Plan:
- Ran `bin/repository update ...` on an observed, bare repository.
- ...on an observed, non-bare ("legacy") repository.
- ...on a hosted, bare repository.
Maniphest Tasks: T13479
Differential Revision: https://secure.phabricator.com/D20945
Summary: Fixes T13476. Policy tags in object headers and "Visible To" controls in some dialog contexts may stack and wrap oddly. Improve spacing so they don't overlap visually when wrapping.
Test Plan: Viewed affected interfaces in narrow and wide windows.
Maniphest Tasks: T13476
Differential Revision: https://secure.phabricator.com/D20944
Summary:
Fixes T13471. Recent versions of PHP raise a warning when this function is called.
We're only calling it so we can instantly fatal if it's enabled, so use "@" to silence the warning.
Test Plan: Loaded site; see also T13471 for a user reporting that this fix is effective.
Maniphest Tasks: T13471
Differential Revision: https://secure.phabricator.com/D20942
Summary: Fixes T13472. This library uses `$a{0}`, but this is deprecated in favor of `$a[0]`.
Test Plan:
Ran `bin/search index Txxx --force` on a task with "filing" in the title (this term reaches the "m" rule of the stemmer).
(I'm not on new enough PHP for this to actually raise an error, but I'll follow up with the reporting user.)
Maniphest Tasks: T13472
Differential Revision: https://secure.phabricator.com/D20941
Summary: Ref T13472. Ref T13395. These classes are only used by Phabricator and not likely to find much use in Arcanist.
Test Plan: Grepped libphutil and Arcanist for removed symbols.
Maniphest Tasks: T13472, T13395
Differential Revision: https://secure.phabricator.com/D20939
Summary:
See <https://hackerone.com/reports/758002>. The link rules don't test that their parameters are flat text before using them in unsafe contexts.
Since almost all rules are lower-priority than these link rules, this behavior isn't obvious. However, two rules have broadly higher priority (monospaced text, and one variation of link rules has higher priority than the other), and the latter can be used to perform an XSS attack with input in the general form `()[ [[ ... | ... ]] ]` so that the inner link rule is evaluated first, then the outer link rule uses non-flat text in an unsafe way.
Test Plan:
Tested examples in HackerOne report. A simple example of broken (but not unsafe) behavior is:
```
[[ `x` | `y` ]]
```
Differential Revision: https://secure.phabricator.com/D20937
Summary: Maniphest object has `getURI` method, let's use it
Test Plan: Create event in task - URI generated as expected in email notification
Reviewers: epriestley, Pawka, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin
Differential Revision: https://secure.phabricator.com/D20935
Summary:
Ref T13362. Some applications moved to fixed-width a while ago but I was generally unsatisfied with where they ended up and have been pushing them back to full-width.
Push Config back to full-width. Some of the subpages end up a little weird, but this provides more space to work with to make some improvements, like makign `maniphest.statuses` more legible in the UI>
Test Plan: Grepped for `setFixed(`, updated each page in `/config/`. Browsed each controller, saw workable full-width UIs.
Maniphest Tasks: T13362
Differential Revision: https://secure.phabricator.com/D20925
Summary:
See PHI1558. Ref T11860. Ref T13444. I partly implemented PHIDs for "UserEmail" objects, but they can't load on their own so you can't directly `bin/remove destroy` them yet.
Allow them to actually load by implementing "PolicyInterface".
Addresses are viewable and editable by the associated user, unless they are a bot/list address, in which case they are viewable and editable by administrators (in preparation for T11860). This has no real impact on anything today.
Test Plan:
- Used `bin/remove destroy <phid>` to destroy an individual email address.
- Before: error while loading the object by PHID in the query policy layer.
- After: clean load and destroy.
Maniphest Tasks: T13444, T11860
Differential Revision: https://secure.phabricator.com/D20927
Summary: Fixes T13465. This "phlog()" made some degree of sense at one time, but is no longer useful or consistent. Get rid of it. See T13465 for discussion.
Test Plan: Made a conduit call that hit a policy error, no longer saw error in log.
Maniphest Tasks: T13465
Differential Revision: https://secure.phabricator.com/D20924
Summary:
Fixes T13464. Fully-realized paths have a "path" (normalized, effective path) and a "display" path (user-facing, un-normalized path).
During transaction validation we build ref keys for paths before we normalize the "display" values. A few different approaches could be taken to resolve this, but just default the "display" path to the raw "path" if it isn't present since that seems simplest.
Test Plan: Edited paths in an Owners package, no longer saw a warning in the logs.
Maniphest Tasks: T13464
Differential Revision: https://secure.phabricator.com/D20923
Summary: Ref T13444. Allow the effects of performing an identity rebuild to be previewed without committing to any changes.
Test Plan: Ran "bin/repository rebuild-identities --all-identities" with and without "--dry-run".
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20922
Summary:
Fixes T13457. Ref T13444. When we iterate over commits in a particular repository, the default iteration strategy can't effectively use the keys on the table.
Tweak the ordering so the "<repositoryID, epoch, [id]>" key can be used.
Test Plan:
- Ran `bin/audit delete --repository X` and `bin/repository rebuild-identities --repository X` before and after changes.
- With just the key changes, performance was slightly better. My local data isn't large enough to really emphasize the key changes.
- With the page size changes, performance was a bit better (~30%, but on 1-3 second run durations).
- Used `--trace` and ran `EXPLAIN ...` on the new queries, saw them select the "<repositoryID, epoch, [id]>" key and report a bare "Using index condition" in the "Extra" column.
Maniphest Tasks: T13457, T13444
Differential Revision: https://secure.phabricator.com/D20921
Summary:
Ref T13444. Currently, many mutations to users and email addresses (particularly: user creation; and user and address destruction) do not propagate properly to repository identities.
Add hooks to all mutation workflows so repository identities get rebuilt properly when users are created, email addresses are removed, users or email addresses are destroyed, or email addresses are reassigned.
Test Plan:
- Added random email address to account, removed it.
- Added unassociated email address to account, saw identity update (and associate).
- Removed it, saw identity update (and disassociate).
- Registered an account with an unassociated email address, saw identity update (and associate).
- Destroyed the account, saw identity update (and disassociate).
- Added address X to account A, unverified.
- Invited address X.
- Clicked invite link as account B.
- Confirmed desire to steal address.
- Saw identity update and reassociate.
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20914
Summary:
Ref T13444. To interact meaningfully with "DestructionEngine", objects need a PHID. The "UserEmail" object currently does not have one (or a real "Query").
Provide basic PHID support so "DestructionEngine" can interact with the object more powerfully.
Test Plan:
- Ran migrations, checked data in database, saw sensible PHIDs assigned.
- Added a new email address to my account, saw it get a PHID.
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20913
Summary: Ref T13444. Prepare to hook identity updates when user email addreses are destroyed.
Test Plan:
- Destroyed a user with `bin/remove destroy ... --trace`, saw email deleted.
- Destroyed an email from the web UI, saw email deleted.
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20912
Summary:
Ref T13444. Repository identities have, at a minimum, some bugs where they do not update relationships properly after many types of email address changes.
It is currently very difficult to fix this once the damage is done since there's no good way to inspect or rebuild them.
Take some steps toward improving observability and providing repair tools: allow `bin/repository rebuild-identities` to effect more repairs and operate on identities more surgically.
Test Plan: Ran `bin/repository rebuild-identities` with all new flags, saw what looked like reasonable rebuilds occur.
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20911
Summary: Ref T13444. Send all repository identity/detection through a new "DiffusionRepositoryIdentityEngine" which handles resolution and detection updates in one place.
Test Plan:
- Ran `bin/repository reparse --message ...`, saw author/committer identity updates.
- Added "goose@example.com" to my email addresses, ran daemons, saw the identity relationship get picked up.
- Ran `bin/repository rebuild-identities ...`, saw sensible rebuilds.
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20910
Summary: Ref T13444. This is an ancient event and part of the old event system. It is not likely to be in use anymore, and repository identities should generally replace it nowadays anyway.
Test Plan: Grepped for constant and related methods, no longer found any hits.
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20909
Summary:
Ref T13444. You can currently explicitly unassign an identity (useful if the matching algorithm is misfiring). However, this populates the "currentEffectiveUserPHID" with the "unassigned()" token, which mostly makes things more difficult.
When an identity is explicitly unassigned, convert that into an explicit `null` in the effective user PHID.
Then, realign "assigned" / "effective" language a bit. Previously, `withAssigneePHIDs(...)` actualy queried effective users, which was misleading. Finally, bulk up the list view a little bit to make testing slightly easier.
Test Plan:
- Unassigned an identity, ran migration, saw `currentEffectiveUserPHID` become `NULL` for the identity.
- Unassigned a fresh identity, saw NULL.
- Queried for various identities under the modified constraints.
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20908
Summary:
Ref T13444. Currently, identities for a particular email address are queried with "LIKE" against a binary column, which makes the query case-sensitive.
- Extract the email address into a separate "sort255" column.
- Add a key for it.
- Make the query a standard "IN (%Ls)" query.
- Deal with weird cases where an email address is 10000 bytes long or full of binary junk.
Test Plan:
- Ran migration, inspected database for general sanity.
- Ran query script in T13444, saw it return the same hits for "git@" and "GIT@".
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T13444
Differential Revision: https://secure.phabricator.com/D20907
Summary:
Depends on D20919. Ref T13462. When editing milestones, we currently predict they will have no members for policy evaluation purposes. This isn't the right rule.
Instead, predict that their membership will be the same as the parent project's membership, and pass this hint to the policy layer.
See T13462 for additional context and discussion.
Test Plan:
- Set project A's edit policy to "Project Members".
- Joined project A.
- Tried to create a milestone of project A.
- Before: policy exception that the edit policy excludes me.
- After: clean milestone creation.
- As a non-member, tried to create a milestone. Received appropriate policy error.
Maniphest Tasks: T13462
Differential Revision: https://secure.phabricator.com/D20920
Summary:
Ref T13462. Currently, when testing milestone edit policies during creation, the project object does not behave like a milestone:
- it doesn't have a milestone number yet, so it doesn't try to access the parent project; and
- the parent project isn't attached yet.
Instead: attach the parent project sooner (which "should" be harmless, although it's possible this has weird side effects); and give the adjusted policy object a dummy milestone number if it doesn't have one yet. This forces it to act like a milestone when emitting policies.
Test Plan:
- Set "Projects" application default edit policy to "No One".
- Created a milestone I had permission to create.
- Before: failed with a policy error, because the project behaved like a non-milestone and returned "No One" as the effective edit policy.
- After: worked properly, correctly evaluting the parent project edit policy as the effective edit policy.
- Tried to create a milestone I did not have permission to create (no edit permission on parent project).
- Got an appropriate edit policy error.
Maniphest Tasks: T13462
Differential Revision: https://secure.phabricator.com/D20919
Summary:
Fixes T13461. Some applications provide hints about policy strength in the header, but these hints are inconsistent and somewhat confusing. They don't make much sense for modern objects with Custom Forms, which don't have a single "default" policy.
Remove this feature since it seems to be confusing things more than illuminating them.
Test Plan:
- Viewed various objects, no longer saw colored policy hints.
- Grepped for all removed symbols.
Maniphest Tasks: T13461
Differential Revision: https://secure.phabricator.com/D20918
Summary: Fixes T13456. These edits are remarkup edits and should attach files, trigger mentions, and so on.
Test Plan: Created a text panel, dropped a file in. After changes, saw the file attach properly.
Maniphest Tasks: T13456
Differential Revision: https://secure.phabricator.com/D20906
Summary:
Ref T13454. Fixes T13006. When a user provide us with an SSH private key and (possibly) a passphrase:
# Try to verify that they're correct by extracting the public key.
# If that fails, try to figure out why it didn't work.
Our success in step (2) will vary depending on what the problem is, and we may end up falling through to a very generic error, but the outcome should generally be better than the old approach.
Previously, we had a very unsophisticated test for the text "ENCRYPTED" in the key body and questionable handling of the results: for example, providing a passphrase when a key did not require one did not raise an error.
Test Plan:
Created and edited credentials with:
- Valid, passphrase-free keys.
- Valid, passphrased keys with the right passphrase.
- Valid, passphrase-free keys with a passphrase ("surplus passphrase" error).
- Valid, passphrased keys with no passphrase ("missing passphrase" error).
- Valid, passphrased keys with an invalid passphrase ("invalid passphrase" error).
- Invalid keys ("format" error).
The precision of these errors will vary depending on how helpful "ssh-keygen" is.
Maniphest Tasks: T13454, T13006
Differential Revision: https://secure.phabricator.com/D20905
Summary:
Fixes T13443. When a panel raises an exception during edit action generation, it currently escapes to top level. Instead, catch it more narrowly.
Additionally, mark "ChangesetSearchEngine" as not being a suitable search engine for use in query panels. There's no list view or search URI so it can't generate a sensible panel.
Test Plan:
- Added a changeset panel to a dashboard.
- Before: entire dashboard fataled.
- After: panel fataled narrowly, menu fatals narrowly, dashboard no longer permits creation of another Changeset query panel.
Maniphest Tasks: T13443
Differential Revision: https://secure.phabricator.com/D20902
Summary:
Ref T13442. Ref T13157. There's a secret URI to look at an object's hovercard in a standalone view, but it's hard to remember and impossible to discover.
In developer mode, add an action to "View Hovercard". Also add "View Handle", which primarily shows the object PHID.
Test Plan: Viewed some objects, saw "Advanced/Developer...". Used "View Hovercard" to view hovercards and "View Handle" to view handles.
Maniphest Tasks: T13442, T13157
Differential Revision: https://secure.phabricator.com/D20887
Summary:
Ref T13453. Some of the Asana integrations also need API updates.
Depends on D20899.
Test Plan:
- Viewed "asana.workspace-id" in Config, got a sensible GID list.
- Created a revision, saw the associated Asana task get assigned.
- Pasted an Asana link I could view into a revision description, saw it Doorkeeper in the metadata.
Maniphest Tasks: T13453
Differential Revision: https://secure.phabricator.com/D20900
Summary: Ref T13453. The Asana API has changed, replacing all "id" fields with "gid", including the "users/me" API call result.
Test Plan: Linked an Asana account. Before: error about missing 'id'. After: clean link.
Maniphest Tasks: T13453
Differential Revision: https://secure.phabricator.com/D20899
Summary:
Fixes T13452. We currently give users mixed signals about the interaction mode of this text: the cursor says "text" but the behavior is "grab".
Make the behavior "text" to align with the cursor. An alternate variation of this change is to remove the cursor, but this is preferable if it doesn't cause problems, since copying the task ID is at least somewhat useful.
Test Plan: In Safari, Firefox, and Chrome: selected and copied object names from workboard cards; and dragged workboard cards by other parts of their UI.
Maniphest Tasks: T13452
Differential Revision: https://secure.phabricator.com/D20898
Summary: Ref T13425. When a change adds or removes a block (vs adding or removing a document, or changing a block), we currently try to render `null` as cell content in the unified view. Make this check broader to catch both "no opposite document" and "no opposite cell".
Test Plan: {F7008772}
Subscribers: artms
Maniphest Tasks: T13425
Differential Revision: https://secure.phabricator.com/D20897
Summary: Ref T13448. Minor UI issue: setting a "Fetch Refs" value does not highlight the associated menu item, but should.
Test Plan: Set only "Fetch Refs", now saw menu item highlighted.
Maniphest Tasks: T13448
Differential Revision: https://secure.phabricator.com/D20895
Summary:
Fixes T13448. We currently "git clone" to initialize repositories, but this will fetch too many refs if "Fetch Refs" is configured.
In modern Phabricator, there's no apparent reason to "git clone"; we can just "git init" instead. This workflow naturally falls through to an update, where we'll do a "git fetch" and pull in exactly the refs we want.
Test Plan:
- Configured an observed repository with "Fetch Refs".
- Destroyed the working copy.
- Ran "bin/repository pull X --trace --verbose".
- Before: saw "git clone" pull in the world.
- After: saw "git init" create a bare empty working copy, then "git fetch" fill it surgically.
Both flows end up in the same place, this one is just simpler and does less work.
Maniphest Tasks: T13448
Differential Revision: https://secure.phabricator.com/D20894
Summary:
Ref T13448. The default behavior of "git fetch '+refs/heads/master:refs/heads/master'" is to follow and fetch associated tags.
We don't want this when we pass a narrow refspec because of "Fetch Refs" configuration. Tell Git that we only want the refs we explicitly passed.
Note that this doesn't prevent us from fetching tags (if the refspec specifies them), it just stops us from fetching extra tags that aren't part of the refspec.
Test Plan:
- Ran "bin/repository pull X --trace --verbose" in a repository with a "Fetch Refs" configuration, saw only "master" get fetched (previously: "master" and reachable tags).
- Ran "git fetch --no-tags '+refs/*:refs/*'", saw tags fetched normally ("--no-tags" does not disable fetching tags).
Maniphest Tasks: T13448
Differential Revision: https://secure.phabricator.com/D20893
Summary:
Ref T13448. When "Fetch Refs" is configured:
- We switch to a narrow mode when running "ls-remote" against the local working copy. This excludes surplus refs, so we'll incorrectly detect that the local and remote working copies are identical in cases where the local working copy really has surplus refs.
- We rely on "--prune" to remove surplus local refs, but it only prunes refs matching the refspecs we pass "git fetch". Since these refspecs are very narrow under "Fetch Only", the pruning behavior is also very narrow.
Instead:
- When listing local refs, always list everything. If we have too much stuff locally, we want to get rid of it.
- When we identify surplus local refs, explicitly delete them instead of relying on "--prune". We can just do this in all cases so we don't have separate "--prune" and "manual" cases.
Test Plan:
- Created a new repository, observed from a GitHub repository, with many tags/refs/branches. Pulled it.
- Observed lots of refs in `git for-each-ref`.
- Changed "Fetch Refs" to "refs/heads/master".
- Ran `bin/repository pull X --trace --verbose`.
On deciding to do something:
- Before: since "master" did not change, the pull declined to act.
- After: the pull detected surplus refs and deleted them. Since the states then matched, it declined further action.
On pruning:
- Before: if the pull was forced to act, it ran "fetch --prune" with a narrow refspec, which did not prune the working copy.
- After: saw working copy pruned explicitly with "update-ref -d" commands.
Also, set "Fetch Refs" back to the default (empty) and pulled, saw everything pull.
Maniphest Tasks: T13448
Differential Revision: https://secure.phabricator.com/D20892
