Summary:
Ref T5955. Summary of intended changes:
**Improve Granularity of Authorization**: Currently, users have one Conduit Certificate. This isn't very flexible, and means that you can't ever generate an API token with limited permissions or IP block controls (see T6706). This moves toward a world where you can generate multiple tokens, revoke them individually, and assign disparate privileges to them.
**Standardize Token Management**: This moves Conduit to work the same way that sessions, OAuth authorizations, and temporary tokens already work, instead of being this crazy bizarre mess.
**Make Authentication Faster**: Authentication currently requires a handshake (conduit.connect) to establish a session, like the web UI. This is unnecessary from a security point of view and puts an extra round trip in front of all Conduit activity. Essentially no other API anywhere works like this.
**Make Authentication Simpler**: The handshake is complex, and involves deriving hashes. The session is also complex, and creates issues like T4377. Handshake and session management require different inputs.
**Make Token Management Simpler**: The certificate is this huge long thing right now, which is not necessary from a security perspective. There are separate Arcanist handshake tokens, but they have a different set of issues. We can move forward to a token management world where neither of these problems exist.
**Lower Protocol Barrier**: The simplest possible API client is very complex right now. It should be `curl`. Simplifying authentication is a necessary step toward this.
**Unblock T2783**: T2783 is blocked on nodes in the cluster making authenticated API calls to other nodes. This provides a simpler way forward than the handshake mess (or enormous-hack-mess) which would currently be required.
Test Plan:
- Generated tokens.
- Generated tokens for a bot account.
- Terminated tokens (and for a bot account).
- Terminated all tokens (and for a bot account).
- Ran GC and saw it reap all the expired tokens.
NOTE: These tokens can not actually be used to authenticate yet!
{F249658}
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T5955
Differential Revision: https://secure.phabricator.com/D10985
Summary:
Two goals:
- If an attacker compromises an administrator account (without compromising the host itself), they can currently take advantage of vulnerabilities in prototype applications by enabling the applications, then exploiting the vulnerability. Locking this option requires CLI access to enable prototypes, so installs which do not have prototypes enabled have no exposure to security issues in prototype applications.
- Making this very slightly harder to enable is probably a good thing, given the state of the world and support.
Test Plan: Verified that web UI shows the value is locked and instructs the user to update via the CLI.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10993
Summary: Via HackerOne. An attacker with access to both Phame and the filesystem could potentially load a skin that lives outside of the configured skin directories, because we had insufficient checks on the actual skin at load time.
Test Plan: Attempted to build a blog with an invalid skin; got an exception instead of a mis-load of a sketchy skin.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10992
Summary: Fixes T6748. This just didn't get aligned when CCs became a modern transaction.
Test Plan: Added a CC to a task, used `bin/mail show-outbound` to verify it showed up as a CC tag.
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Maniphest Tasks: T6748
Differential Revision: https://secure.phabricator.com/D10991
Summary:
Ref T2783. This is primarily exploratory and just figuring out what we're blocked on:
- Allow a Repository to be bound to a Service. The Service may eventually define multiple read/write nodes, etc.
- There's no UI to do this binding yet, you have to touch the database manually.
- If a repository is bound to a Service, effect Conduit calls via calls to the remote service instead of executing them in-process.
- These don't actually work yet since there's no authentication (see T5955).
Test Plan:
- Made a nice Service with a nice Binding to a nice Interface on a nice Device.
- Force-associated a repository with the service using a raw MySQL query.
- Saw Phabricator try to make a remote call to the service (on localhost) and fail because of missing auth stuff.
- Also ran `almanac.queryservices`.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2783
Differential Revision: https://secure.phabricator.com/D10982
Summary: Fixes T6595. This diff has two issues as is... 1) the differential data fetching is pretty cheesey, but it looks like we can't just issue three separate databases to get the right data? 2) the translations break, since I am turning this into a string (and not an int) so the whole pluralization bit fails. I think 1 is okay as is and 2 needs to be fixed though I am not sure how to best do that...
Test Plan: loaded home page and it looked nice...!
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6595
Differential Revision: https://secure.phabricator.com/D10979
Summary:
Fixes T6651, T6682. Since policy is defined by ancestry, you can't make things outside the core tree.
An alternative fix would be to automagically stub everything in these cases. This has potential negative policy implications - consider making a public document with several levels of depth that automagically stubs out its ancestry as public - and additionally the PhabricatorApplicationTransactionEditor framework would make this very tricky code (i.e. you are expected to validateTransactions in said hook *and* return an error if things aren't valid and not do some automagic stubbing, etc.)
Test Plan: tried to move a doc from location/that/exists to locationz/thatz/dontz/existz/ and got an error message with links to each missing doc. tried to create a doc at locatonz/thatz/dontz/existsz/ and got an error message with links to each missing doc.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6682, T6651
Differential Revision: https://secure.phabricator.com/D10978
Summary: (some) international keyboard layouts can not type "~" in a way as to trigger this so use "@" instead. Save the suggested "+" as that seems like it would be useful for some future "adding stuff" keyboard workflow. Pretty stoked to get this squared away as I am quite confident our unreleased product will now be a huge smashing success. Ref T6683.
Test Plan: made sure my choice was okay via https://en.wikipedia.org/wiki/Dead_key#Dead_keys_on_various_keyboard_layouts; used the "@" key to show all transactions
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6683
Differential Revision: https://secure.phabricator.com/D10983
Summary: Fixes T6734. This is a very generic fix, which basically attaches the subscribers if necessary. This seems like a good idea given there's some crazy generic code doing this sort of thing? This would end up being a new pattern for these types of objects that may be loaded by a general object query but then get some editor action against them.
Test Plan: I can't actually reproduce this in my sandbox so I'll verify live again.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6734
Differential Revision: https://secure.phabricator.com/D10976
Summary:
There's a comma to the lower-left of my profile picture here:
{F248962}
This is on a page like https://secure.phabricator.com/F248948
What's happening is that some `render()` method is returning a valid result like `array($stuff, null)`. This is getting passed to JS as an array, which is implicitly `join()`'ing it into a string, adding a comma.
Instead, make sure we render these to strings on the server side before shipping them to the client.
Test Plan: No more comma on file previews.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10974
Summary: Ref T5604. Found this trying to open T5604 live. Basically this internal query needs the needSubscriberPHID set to true.
Test Plan: doing it live
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T5604
Differential Revision: https://secure.phabricator.com/D10975
Summary: Fixes T6731. I don't really understand the intent behind the two view classes here, but to get this to work I need to pass yet more data to the lower-level class.
Test Plan: Viewed a task with many comments. Clicked "show older". Quoted everything I could. Verified for each quote that it quoted correctly, inlcuding linking to the prior transaction.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6731
Differential Revision: https://secure.phabricator.com/D10973
Summary: (Needed a clean branch). Moves the field up and renames to Query
Test Plan: Visit Maniphest Search, see new field, test a query
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10971
Summary: Some browsers, mobile, didn't show 1px space between objects (due to font renderings). This enforces the size for consistent cross-browser display.
Test Plan: Test IE10
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10970
Summary: Fixes T6727. Repro is: mention a task on another task, in a comment.
The inverse edge editor applying the "alincoln mentioned this in <other task>" transaction doesn't have enough data to execute Herald rules.
Just don't try to execute the rules, since they don't make much sesne from a product perspective and are tricky from a technical perspective.
Test Plan: Commented on `T1` with `T2` in comment body and a Herald rule that examines subscribers.
Reviewers: btrahan
NOTE: Cowboy committing this since any task mention fatals.
Summary: Fixes T5604. This should fix some random bugs, lets us move forward more easily, and all that good stuff about killing code debt.
Test Plan:
- Conduit method maniphest.createtask
- verified creating user subscribed
- verified subscription transaction
- Conduit method maniphest.update
- verified subscribers set as specified to ccPHIDs parameter
- verified subscription transaction
- Herald
- verified herald rule to add subscriber worked
- verified no subscribers removed accidentally
- edit controller
- test create and verify author gets added IFF they put themselves in subscribers control box
- test update gets set to exactly what user enters
- lipsum generator'd tasks work
- bulk add subscribers works
- bulk remove subscriber works
- detail controller
- added myself by leaving a comment
- added another user via explicit action
- added another user via implicit mention
- task merge via search attach controller
- mail reply handler
- add subscriber via ./bin/mail receive-test
- unsubscribe via ./bin/mail receive-test
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T5604
Differential Revision: https://secure.phabricator.com/D10965
Summary: I didn't get this quite right.
Test Plan:
- Clicked to open, saw white, then closed by:
- Clicking document outside menu;
- clicking menu icon again;
- clicking a different menu icon.
- In all three cases, got correct close + un-white behavior.
Reviewers: chad
Reviewed By: chad
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10967
Summary: I think this is what you're after?
Test Plan: clicky clicky
Reviewers: chad
Reviewed By: chad
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10966
Summary: Fixes T6598, "Primary Hashtag" field should only be visible in edit mode of existing projects.
Test Plan: Create project, "Primary Hashtag" field should be hidden. Edit an existing project, "Primary Hashtag" field should appear above "Additional Hashtags" as before.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin, epriestley
Maniphest Tasks: T6598
Differential Revision: https://secure.phabricator.com/D10964
Summary:
Ref T2783. ConduitCall currently has logic to pick a random remote server, but this is ultimately not appropriate: we always want to send requests to a specific server. For example, we want to send repository requests to a server which has that repository locally. The repository tier is not homogenous, so we can't do this below the call level.
Make ConduitCall always-local; logic above it will select ConduitCall for an in-process request or do service selection for an off-host request via ConduitClient.
Test Plan:
- Browsed some pages using ConduitCall, everything worked.
- Grepped for removed stuff.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2783
Differential Revision: https://secure.phabricator.com/D10959
Summary: Fixes T6683.
Test Plan: clicked the yellow box and it worked! pressed '~' and it worked!
Reviewers: epriestley, chad
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6683
Differential Revision: https://secure.phabricator.com/D10932
Summary: Hides the second column of info on narrow (33%) columns on dashboards on object-items.
Test Plan:
Tested a narrow left and narrow right column, 160px is gained back.
{F248276}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10962
Summary: This adds back the title to the header link and scans through the codebase for instances where
Test Plan: Tested as many ObjectItemLists as I could find (each app homepage), there may be outliers, but can resolve those individually.
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10961
Summary: Ref T6723. This allows you to hover over truncated headers and get the full text if needed.
Test Plan:
Hover over header, see full title
{F248197}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6723
Differential Revision: https://secure.phabricator.com/D10958
Summary: Ref T6723. Because of the sidenav, some dashboard layouts may perform poorly on more narrow desktop displays. This provides an additional media query and CSS rules.
Test Plan:
Move viewport to 1001 px, see desktop, move to 999 pixels, see tablet-like display.
{F248158}
{F248159}
{F248160}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6723
Differential Revision: https://secure.phabricator.com/D10956
Summary: Fixes T6562, Title/Description querying for Passphrase
Test Plan: Open Passphrase, open advanced queries, enter a title and/or description. Search results should show credentials matching the search.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6562
Differential Revision: https://secure.phabricator.com/D10953
Summary: Ref T6713. We were dropping latest transaction ID. This should fix the "easy" part of T6713.
Test Plan: tried to add participants and it worked! (removing participants only allows yourself, but that worked too.)
Reviewers: chad, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6713
Differential Revision: https://secure.phabricator.com/D10952
Summary: These are not used
Test Plan: grep codebase
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10938
Summary: Fixes T6719. At some point, we added automagical subscriptions via @mentions, and these were failing in project descriptions from a lack of an implementation in the editor. Said "implementation" is to do nothing, but it needs to be there nonetheless.
Test Plan: updated a project mentioning someone in the description and it worked. also saw a 'subscriber added' transaction
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6719
Differential Revision: https://secure.phabricator.com/D10951
Summary: Crumbs misalign because the hit area on the icons was 2px too large.
Test Plan: Test on pages with crumbs. IE, FF, Chrome...
Reviewers: epriestley, btrahan
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10949
Summary: Cleans up spacing, updates to fonts instead of images. Fixed some mobile issues.
Test Plan:
Test with and without counts on desktop, tablet, mobile. Test layout in FF, Chrome, IE.
{F246745}
{F246746}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10948
Summary: Fixes T6664, clicking search icon in empty search field should link to advanced search
Test Plan: navigate to home page, click search icon or click into search box and hit enter. Advanced search page should open.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6664
Differential Revision: https://secure.phabricator.com/D10947
Summary:
The `$timeline` variable is undefined. I was seeing the following error in the logs:
```
EXCEPTION: (RuntimeException) Undefined variable: timeline at [<phutil>/src/error/PhutilErrorHandler.php:210]
#0 PhutilErrorHandler::handleError(integer, string, string, integer, array) called at [<phabricator>/src/applications/harbormaster/controller/HarbormasterStepEditController.php:205]
#1 HarbormasterStepEditController::processRequest() called at [<phabricator>/src/aphront/AphrontController.php:33]
#2 AphrontController::handleRequest(AphrontRequest) called at [<phabricator>/webroot/index.php:103]
```
Test Plan: Created a build step without a fatal error.
Reviewers: btrahan, hach-que, epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10941
Summary: This rule is making text actually more difficult to read on Windows, as well as is broken on Win/Chrome, and evidentally has performance issues on mobile browsers.
Test Plan: Turn it off, can still read web page
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10940
Summary:
Fixes T6702. Ref T3554. Currently, tasks can be cancelled, retried and freed from the web UI by any logged in user.
This isn't appreciably dangerous (I can't come up with a way that a user could do anything security-affecting), but I think I probably intended this to be admin-only, but these actions should move to the CLI anyway.
Move them to the CLI. Lay some groundwork for some future `bin/worker cancel --class SomeTaskClass`, but don't implement that yet.
Test Plan: Used `cancel`, `retry` and `free` from the CLI. Hit all the error/success states.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T3554, T6702
Differential Revision: https://secure.phabricator.com/D10939
Summary: Ref T3669. Probably. Adds a yellow warning at the top of the Diff View and makes the comment draft icon yellow on lists of revisions.
Test Plan:
Test a diff with many warnings, see warning. Test a diff with draft comments, see warning. Test new icon in list view.
{F230133}
{F230134}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T3669
Differential Revision: https://secure.phabricator.com/D10789
Summary: Lots of minor spacing/alignment tweaks on mobile menus
Test Plan: set browser to 320 width, inspect for issues and alignments
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10937
Summary: Fixes T6699. We need to "loadInlineComments" consistently, though for an unexpected reason - this mutates the $changesets to include all $changesets that have an associated inline comment, which is necessary to make them render properly.
Test Plan: Took a diff with inline comments and updated it, noting the inline comments disappeared. applied this patch and the inlines reappeared.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6699
Differential Revision: https://secure.phabricator.com/D10935
Summary: Change icon for Settings app to more match previous. Also align plus icon a little better.
Test Plan: Lots of staring.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10934
Summary: Fixes T6694. Ref T4712. Turns out the logic here was slightly incorrect; we don't want to use the id of the last thing we hid but rather the first thing we show. I had garbage test data ("asdsadsadsa", etc) I guess so I didn't notice this.
Test Plan: made a new task where user a and user b alternated 3 comments each, cooperatively numbering them from 1 - 20. as both users, showed older transactions. pre-patch the issue described in T6694 occurred and post patch I saw the entire counting sequence.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T4712, T6694
Differential Revision: https://secure.phabricator.com/D10933
Summary: we still need to be pager-sensitive, but otherwise this "show all" stuff is dead, dead dead...! Ref T4712. I think we can close the book on T4712 with one more diff to clean up the array_reverse / reverse paging stuff? That diff is probably a bit tricky as it involes auditing every TransactionQuery callsite...
Test Plan: viewed a task with a lot of transactions. clicked "show older" and it worked!
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T4712
Differential Revision: https://secure.phabricator.com/D10926
Summary: Fixes T6693.
Test Plan:
Made a bunch of comments on a diff with differential, being sure to leave inlines here and there. This reproduced the issue in T6693. With this patch this issue no longer reproduces!
Successfully "showed older changes" in Maniphest too.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6693
Differential Revision: https://secure.phabricator.com/D10931
Summary: Updates header to use font-icons instead of images.
Test Plan: Test desktop and mobile layouts, Chrome, FF, Safari, IE.
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10930
Summary: Only necessary for edits, only bother if the comment version is greater than 1. Ref T6690. This is another way to fix T6690 -- this check will never run since you can't edit a conpherence comment -- **but** the fix already applied should happen too to future proof Conpherence.
Test Plan: made a comment on a diff - success. edited the comment and mentions were generated.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6690
Differential Revision: https://secure.phabricator.com/D10928
Summary: Fixes T6690. The editor innards end up loading the conpherence object, whose policy is dictated by these participation objects.
Test Plan: pre patch could not create new conpherences. post patch I can create conpherences! i can also add people to conpherences and it works.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6690
Differential Revision: https://secure.phabricator.com/D10927
