Summary:
Ref T13513. The on-hover-inline reticle has switched over to have cell-based behavior. Switch the on-hover-line-number reticle to use the same behavior.
Also, clean up the dirty/redraw loop slightly: we no longer need to dirty on resize, and we don't need to redraw if the range isn't actually dirty.
Test Plan: Highlighted lines and line ranges. Hovered over inlines.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21262
Summary:
Ref T13513. See PHI1734, which raises a concern about the performance of large revisions near the 100-change threshold.
Currently, `getInlines()` is called whenever the scroll position transitions between two changesets, and it performs a relatively complicated DOM scan to lift inlines out of the document.
This shows up as taking a small but nontrivial amount of time in Firefox profiles and should be safely memoizable.
Test Plan:
- Under Firefox profiling, scrolled through a large revision.
- Before change: `getInlines()` appeared as the highest-cost thing we're explicitly doing on profiles.
- After change: `getInlines()` was no longer meaningfully represented on profiles.
- Created inlines, edited inlines, etc. Didn't identify any broken behavior.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21261
Summary:
Ref T13513. The way I'm highlighting lines won't work for Jupyter notebooks or other complex content blocks, and I don't see an obvious way to make it work that's reasonably robust.
However, we can just ignore the range behavior for complex content and treat the entire block as selected. This isn't quite as fancy as the source behavior, but pretty good.
Also, adjust unified diff behavior to work correctly with highlighting and range selection.
Test Plan:
- Used range selection in a Jupyter notebook, got reasonable behavior (range is treated as "entire block").
- Used range selection in a unified diff, got equivalent behavior to 2-up diffs.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21257
Summary:
Ref T13513. Give selected inlines a selection state and visual cues which are similar to the changeset selection state.
Also fix a couple of minor issues with select interactions and offset comments.
Test Plan: Selected inlines, saw obvious visual cues.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21256
Summary: Ref T13513. Currently, clicking inline or changeset headers eats the click events. This doesn't serve any clear purpose, and means these clicks do not clear text selections from the document, which is unusual.
Test Plan:
- Selected some text in a diff.
- Clicked a changeset header to select it.
- Before patch: text selection and context menu were retained.
- After patch: text selection was cleared and context menu was dismissed.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21255
Summary: Ref T13513. This fixes a bug where clicking a line number, then clicking "Cancel" causes the paths panel to briefly update with an extra inline comment counted on the file.
Test Plan:
- Clicked a line number.
- Typed some text.
- Clicked "Cancel".
- Before patch: paths panel flashes "1".
- After patch: paths panel stays stable.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21253
Summary:
Ref T13513. Currently, when creating an inline by selecting a line range, slightly careless handling leads to an inline with "0" offsets (by passing "undefined" to the server). This causes the block to highlight every line except the last one as fully bright, which is incorrect.
An inline with "0" offsets and an inline with no offsets are different. Be more careful about passing offsets around and rendering them.
Test Plan:
- Used the line numbers to add an inline to lines 4-8 of a change.
- Hovered the inline.
- Saw all four lines marked as "dull"-highlighted (previously: three bright lines, one dull line).
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21252
Summary:
Ref T13513. When a user selects a text range and uses "New Inline Comment" to create a comment directly from a range, store the offset information alongside the comment.
When hovering the comment, highlight the original range.
Test Plan: {F7480926, size=full}
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21250
Summary:
Ref T13513.
- Firefox represents multiple selected rows as a discontinuous range. Accommodate this.
- Unified diffs don't have a "copy" marker. Do something sort-of-reasonable for them.
Test Plan:
- Selected multiple lines of content in Firefox, got an option to add a comment.
- Selected content in unified mode, got an option to add a comment.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21249
Summary: Ref T13513. Support direct text selection for inlines. This is currently just an alternate way to get to the same place as using line numbers, but can preserve offset/range information in the future.
Test Plan:
- Selected some text, hit "c", clicked "New Inline Comment", got sensible comments on both sides of a diff in Safari, Chrome, and (with limitations) Firefox.
- Caveats: no unified support, doesn't work across lines in Firefox.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21248
Summary: Ref T11401. Ref T13513. This paves the way for more comment actions, particularly an edit-after-submit action.
Test Plan: Took all actions from menus, via mouse and via keyboard (where applicable).
Maniphest Tasks: T13513, T11401
Differential Revision: https://secure.phabricator.com/D21244
Summary:
Ref T13513. Currently, "View as Document Type..." lists every available engine.
This is hard to get completely right because we can't always rebuild the document ref accurately in the endpoint, but try harder to fake something reasonable.
Test Plan: Used "View as Document Type..." on Jupyter notebooks, was given "Jupyter" and "Source" as options.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21241
Summary:
Ref T13513. As part of inline metadata, save the document engine the change is being rendered with.
This will allow other parts of the UI to detect that an inline was created on a Jupyter notebook but is being rendered on raw source, or whatever else.
The immediate goal is to fix nonsensical inline snippet rendering in email on Jupyter notebooks.
Test Plan:
- Created inlines and replies on normal soure code, saw no document engine annotated in the database.
- Created inlines and replies on a Jupyter notebook rendered in Jupyter mode, saw "jupyter" annotations in the database.
- Swapped document engines between Jupyter and Source, etc.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21240
Summary: Ref T13513. Currently, if you're editing a comment, "delete" doesn't put the comment into the correct state. This action is normally only reachable from comment previews, since an editing inline has no "delete" button.
Test Plan:
- Started editing an inline, clicked "Delete", got a deletion.
- Created an inline, typed text,
- Deleted a normal comment via preview.
- Deleted a normal comment via the on-inline action.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21238
Summary:
Ref T13513. Currently, clicking "View" from the inline comment preview (below the "add comment" area at the bottom of the page) only works if the inline isn't being edited.
Update this behavior so it works on inlines in either "Viewing" or "Editing" states.
Test Plan:
- Clicked "View" on a normal inline, got jumped/selected.
- Clicked "View" on an editing inline, got jumped/selected.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21237
Summary:
Ref T13513. Currently:
- If you click the "Show Changeset" button, your state change doesn't actually get saved on the server.
- It's hard to select a changeset path name for copy/paste because the "highlight the header" code tends to eat the event.
Instead: persist the former event; make the actual path text not be part of the highlight hitbox.
Test Plan:
- Clicked "Show Changeset", reloaded, saw changeset visibility persisted.
- Selected changeset path text without issues.
- Clicked non-text header area to select/deselect changesets.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21236
Summary:
Ref T13513. Currently, inline storage objects ("TransactionComment") can't directly generate a runtime object ("InlineComment").
Allow this transformation to be performed in a genric way so clunky code which does it per-object-type can be removed, lifted, or simplified.
Simplify an especially gross callsite in preview code.
Test Plan: Previewed inline comments.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21226
Summary:
Ref T13513. Overloading "original text" to get "edit-on-load" comments into the right state has some undesirable side effects.
Instead, provide the text when the editor opens. This fixes a cancel interaction.
Test Plan:
- Create an inline, type text, don't save.
- Reload page.
- Cancel.
- Before: cancelled into empty state.
- After: cancelled into deleted+undo state.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21219
Summary:
Ref T13513. This is a bit clumsy, but the cleanest way to implement "isEditing" inlines today is to send them down as normal inlines and then simulate clicking "edit" on them.
When we do, don't focus the resulting editor: focusing it makes the page scroll around and highlight things in essentially random order as the editors load in.
Test Plan: Reloaded a page with some open editors, wasn't scrolled to them.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21217
Summary:
Ref T13513. As users type text into inline comments, save the comment state as a draft on the server.
This has some rough edges, particularly around previews, but mostly works. See T13513 for notes.
Test Plan: Started an inline, typed some text, waited a second, reloaded the page, saw an editing inline with the saved text.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21216
Summary: Ref T13513. Refine some inline behaviors, see test plan.
Test Plan:
- Edit a comment ("A"), type text ("AB"), cancel, edit.
- Old behavior: edit and undo states (wrong, and undo does not function).
- New behavior: edit state only.
- Edit a comment ("A"), type text ("AB"), cancel. Undo ("AB"), cancel. Edit.
- Old behavior: "AB" (wrong: you never submitted this text).
- New behavior: "A".
- Create a comment, type text, cancel.
- Old behavior: counter appears in filetree (wrong, comment is undo-able but should not be counted).
- New behavior: no counter.
- Cancel editing an empty comment with no text.
- Old behavior: Something buggy -- undo, I think?
- New behavior: it just vanishes (correct behavior).
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21212
Summary:
Ref T13513. Now that the "currently being edited" state of inlines is saved on the server side, clear the flag when the user clicks "Cancel" to leave the "editing" state on the client.
This also serves to delete empty comments.
Test Plan:
- Clicked a line number to create a new comment. Then:
- Clicked "Cancel". Reloaded page, saw no more comment.
- Typed text, saved. Reloaded page, saw non-editing draft. Clicked "Edit", reloaded page, saw editing draft. Clicked "Cancel", reloaded page, saw non-editing draft.
- Typed text, saved. Clicked "Edit", deleted text, saved. Reloaded page, saw no more comment.
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21187
Summary:
Ref T13513. This is mostly an infrastructure cleanup change.
In a perfect world, this would be a series of several changes, but they're tightly interconnected and don't have an obvious clean, nontrivial partition (or, at least, I don't see one). Followup changes will exercise this code repeatedly and all of these individual mutations are "obviously good", so I'm not too worried about the breadth of this change.
---
Inline comments are stored as transaction comments in the `PhabricatorAuditTransactionComment` and `DifferentialTransactionComment` classes.
On top of these two storage classes sit `PhabricatorAuditInlineComment` and `DifferentialInlineComment`. Historically, these were an indirection layer over significantly different storage classes, but nowadays both storage classes look pretty similar and most of the logic is actually the same. Prior to this change, these two classes were about 80% copy/pastes of one another.
Part of the reason they're so copy/pastey is that they implement a parent `Interface`. They are the only classes which implement this interface, and the interface does not provide any correctness guarantees (the storage objects are not actually constrained by it).
To simplify this:
- Make `PhabricatorInlineCommentInterface` an abstract base class instead.
- Lift as much code out of the `Audit` and `Differential` subclasses as possible.
- Delete methods which no longer have callers, or have only trivial callers.
---
Inline comments have two `View` rendering classes, `DetailView` and `EditView`. They share very little code.
Partly, this is because `EditView` does not take an `$inline` object. Historically, it needed to be able to operate on inlines that did not have an ID yet, and even further back in history this was probably just an outgrowth of a simple `<form />`.
These classes can be significantly simplified by passing an `$inline` to the `EditView`, instead of individually setting all the properties on the `View` itself. This allows the `DetailView` and `EditView` classes to share a lot of code.
The `EditView` can not fully render its content. Move the content rendering code into the view.
---
Prior to this change, some operations need to work on inlines that don't have an inline ID yet (we assign an ID the first time you "Save" a comment). Since "editing" comments will now be saved, we can instead create a row immediately.
This means that all the inline code can always rely on having a valid ID to work with, even if that ID corresponds to an empty, draft, "isEditing" comment. This simplifies more code in `EditView` and allows the "create" and "reply" code to be merged in `PhabricatorInlineCommentController`.
---
Client-side inline events are currently handled through a mixture of `ChangesetList` listeners (good) and ad-hoc row-level listeners (less good). In particular, the "save", "cancel", and "undo" events are row-level. All other events are list-level.
Move all events to list-level. This is supported by all inlines now having an ID at all stages of their lifecycle.
This allows some of the client behavior to be simplified. It currently depends on binding complex ad-hoc dictionaries into event handlers in `_drawRows()`, but it seems like almost all of this code can be removed. In fact, no more than one row ever seems to be drawn, so this code can probably be simplified further.
---
Finally, save an "isEditing" state. When we rebuild a revision on the client, click the "edit" button if it's in this state. This is a little hacky, but simpler to get into a stable state, since the row layout of an inline depends on a "view row" followed by an "edit row".
Test Plan:
- Created comments on either side of a diff.
- Edited a comment, reloaded, saw edit stick.
- Saved comments, reloaded, saw save stick.
- Edited a comment, typed text, cancelled, "unedited" to get state back.
- Created a comment, typed text, cancelled, "unedited" to get state back.
- Deleted a comment, "undeleted" to get state back.
Weirdness / known issues:
- Drafts don't autosave yet.
- Fixed in D21187:
- When you create an empty comment then reload, you get an empty editor. This is a bit silly.
- "Cancel" does not save state, but should, once drafts autosave.
- Mostly fixed in D21188:
- "Editing" comments aren't handled specially by the overall submission flow.
- "Editing" comments submitted in that state try to edit themselves again on load, which doesn't work.
Subscribers: jmeador
Maniphest Tasks: T13513
Differential Revision: https://secure.phabricator.com/D21186
Summary:
Ref T13516.
- Add an "Add Comment" navigation anchor.
- Make selection state more clear.
- Make hidden state tidier and more clear.
- Hide "View Options" in the hidden state to dodge all the weird behaviors it implies.
- Click to select/deselect changesets.
- When you open the view dropdown menu, then press "h", close the dropdown menu.
Test Plan: Fiddled with all these behaviors.
Maniphest Tasks: T13516
Differential Revision: https://secure.phabricator.com/D21161
Summary:
Ref T13516. Minor improvements here:
- Show key commands in the "View Options" dropdown.
- Organize it slightly better.
- Improve disabled item behaviors a little bit.
- Add a "Browse Directory" action.
- Rename "...in Diffusion" to "...in Repository".
- Make "d", "D", and "h" use the same targeting rules as "\".
- When you hide a file with the "h" menu item, select it.
Test Plan: Poked at the menu a lot, ran into less questionable behavior.
Maniphest Tasks: T13516
Differential Revision: https://secure.phabricator.com/D21160
Summary:
Ref T13455. Make "hidden" a changeset property similar to other changeset properties.
We don't need to render this on the server, so we make a request (to update the setting) and just discard the response.
Test Plan: {F7375468}
Maniphest Tasks: T13455
Differential Revision: https://secure.phabricator.com/D21158
Summary: Ref T13516. Mark low-importance changes (generated code, deleted files) and owned-with-authority changes in the filetree.
Test Plan: {F7375327}
Maniphest Tasks: T13516
Differential Revision: https://secure.phabricator.com/D21157
Summary:
Ref T13516. Deletes all old filetree / flex / active / collapse nav code in favor of the new code.
Restores the inline tips in the path tree.
Test Plan: {F7374175}
Maniphest Tasks: T13516
Differential Revision: https://secure.phabricator.com/D21154
Summary: Ref T13516. Apply basic UI styling to the new UI and make some more interaction work.
Test Plan: {F7374096}
Maniphest Tasks: T13516
Differential Revision: https://secure.phabricator.com/D21153
Summary: Ref T13516. Generate a tree structure based on the page changesets. Still missing styles and a whole lot of behavior.
Test Plan: {F7373967}
Maniphest Tasks: T13516
Differential Revision: https://secure.phabricator.com/D21152
Summary: Ref T13516. This glues "FormationView" to "ChangesetList". The actual tree is not functional in any meaningful way yet.
Test Plan: {F7373838}
Maniphest Tasks: T13516
Differential Revision: https://secure.phabricator.com/D21151
Summary:
Ref T13515. Currently, "Open in Editor" only works with a file-level selection.
- If we have a change-level or inline-level selection, open the parent changeset.
- If we have no selection, but the banner is showing something, open the fine shown in the banner.
Test Plan: With files, inlines, changes, and no selection, pressed "\". Saw files pop open in my external editor.
Maniphest Tasks: T13515
Differential Revision: https://secure.phabricator.com/D21148
Summary:
Ref T13515. Adding "\" ("Open in External Editor") made this slighlty worse, but it was already pretty bad.
Long ago the keys had a special style on them, but this got changed and dropped somewhere around D16568 -- although at the time, I think they still had a grey background (see T11654).
Some later change removed this background.
Put the background back and separate the keystrokes into groups.
Test Plan: {F7370615}
Maniphest Tasks: T13515
Differential Revision: https://secure.phabricator.com/D21141
Summary:
Ref T13455. Update the other "view state" properties to work like "highlight" now works.
Some complexity here arises from these concerns:
- In "View Standalone", we render the changeset inline. This is useful for debugging/development, and desirable to retain.
- In all other cases, we render the changeset with AJAX.
So the client needs to be able to learn about the "state" properties of the changeset on two different flows. Prior to this change, each pathway had a fair amount of unique code.
Then, some bookkeeping issues:
- At inital rendering time, we may not know which renderer will be selected: it may be based on the client viewport dimensions.
- Prior to this change, the client didn't separate "value of the property for the changeset as rendered" and "desired value of the property".
Test Plan:
- Viewed changes in Differential, Diffusion, and in standalone mode.
- Toggled renderer, character sets, and document engine (this one isn't terribly useful). Reloaded, saw them stick.
- Started typing a comment, cancelled it, hit the undo UI.
Maniphest Tasks: T13455
Differential Revision: https://secure.phabricator.com/D21138
Summary: Ref T13515. See PHI1661. If a file is selected, add a keystroke to click the "Open in External Editor" link.
Test Plan: In Safari, Chrome, and Firefox: used "J" to select a file, then "\" to open it in an external editor. (In Safari and Chrome, this prompts.)
Maniphest Tasks: T13515
Differential Revision: https://secure.phabricator.com/D21135
Summary:
Fixes T13437. This URI construction was just missing URI encoding.
Also, trim the symbol because my test case ended up catching "#define\n" as symbol text.
Test Plan:
- Configured a repository to have PHP symbols.
- Touched a ".php" file with "#define" in it.
- Diffed the change.
- Command-clicked "#define" in the UI, in Safari/MacOS, to jump to the definition.
- Before: taken to a nonsense page where "#define" became an anchor.
- After: taken to symbol search for "#define".
Maniphest Tasks: T13437
Differential Revision: https://secure.phabricator.com/D20876
Summary:
Ref T13425. Currently, code inputs and all outputs are grouped into a single block. This is fine for display notebooks but not great for diffing notebooks.
Instead, split source code input into individual lines with one line per block, and each output into its own block.
This allows you to leave actual line-by-line inlines on source, and comment on outputs individually.
Test Plan: {F6888583}
Maniphest Tasks: T13425
Differential Revision: https://secure.phabricator.com/D20840
Summary:
Depends on D20834. Ref T13425. After the change from "th" to "td" for accessibility, the algorithm picks which cells it should highlight slightly improperly (it picks too many cells since it can no longer find the line numbers).
Ideally, it would probably highlight //only// the source content, but there isn't an easy way to do this right now. Settle for an incremental improvement for the moment.
Test Plan: Hovered over line numbers, saw a more accurate highlight area.
Maniphest Tasks: T13425
Differential Revision: https://secure.phabricator.com/D20835
Summary:
Depends on D20831. Ref T13425. As an escape hatch to get out of future DocumentEngine rendering behavior, provide a "View As.." option.
Now I can break DocumentEngine real bad and no one can complain.
Test Plan: Used "View As" to swap document engines for image files.
Maniphest Tasks: T13425
Differential Revision: https://secure.phabricator.com/D20832
Summary: Ref T13279. Fix some tabular stuff, draw areas better, make the "compose()" API more consistent, unfatal the demo chart, unfatal the project burndown, make the project chart do something roughly physical.
Test Plan: Looked at charts, saw fewer obvious horrors.
Subscribers: yelirekim
Maniphest Tasks: T13279
Differential Revision: https://secure.phabricator.com/D20817
Summary: Ref T13279. We currently draw a point on the chart for each datapoint, but this leads to many overlapping circles. Instead, aggregate the raw points into display points ("events") at the end.
Test Plan: Viewed a stacked area chart with many points, saw a more palatable number of drawn dots.
Subscribers: yelirekim
Maniphest Tasks: T13279
Differential Revision: https://secure.phabricator.com/D20814
Summary:
Fixes T7961. Currently, we present Herald users with actions like "Require legalpad signatures" and "Run build plans" even if Legalpad and Harbormaster are not installed.
Instead, allow fields and actions to be made "unavailable", which means that we won't present them as options when adding to new or existing rules.
If you edit a rule which already uses one of these fields or actions, it isn't affected.
Test Plan:
- Created a rule with a legalpad action, uninstalled legalpad, edited the rule. Action remained untouched.
- Created a new rule, wasn't offered the legalpad action.
- Reinstalled the application, saw the action again.
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T7961
Differential Revision: https://secure.phabricator.com/D20808
Summary:
Ref T4900. When a card is edited, we currently emit an update notification for all the projects the task is tagged with. This isn't quite the right set:
- We want to emit notifications for projects the task //was previously// tagged with, so it can be removed from boards it should no longer be part of.
- We want to emit notifications for ancestors of projects the task is or was tagged with, so parent project boards can be updated.
- However, we don't need to emit notifications for projects that don't actually have workboards.
Adjust the notification set to align better to these rules.
Test Plan:
- Removal of Parent Project: Edited a task on board "A > B", removing the "B" project tag. Saw board A update in another window.
- Normal Update: Edited a task title on board X, saw board X update in another window.
- Used `bin/aphlict debug` to inspect the notification set, saw generally sensible-seeming data going over the wire.
Reviewers: amckinley
Maniphest Tasks: T4900
Differential Revision: https://secure.phabricator.com/D20680
Summary:
Depends on D20654. Ref T4900. When a task is edited, emit a "workboards" event for all boards it appears on (in a future change, this should also include all boards it //previously// appeared on, and all parents of both sets of boards -- but I'm just getting things working for now).
When we receive a "workboards" event, check if the visible board should be updated.
Aphlict has a complicated intra-window leader/follower election system which could let us process this update event exactly once no matter how many windows a user has open with the same workboard. I'm not trying to do any of this since it seems fairly rare. It makes sense for events like "you have new notifications" where we don't want to generate 100 Ajax calls if the user has 100 windows open, but very few users seem likely to have 100 copies of the same workboard open.
Test Plan:
- Ran `bin/aphlict debug`.
- Opened workboard A in two windows, X and Y.
- Edited and moved tasks in window X.
- Saw "workboards" messages in the Aphlict log.
- Saw window Y update in nearly-real-time (locally, this is fast enough that it feels instantaneous).
Then:
- Stopped the Aphlcit server.
- Edited a task.
- Started the Aphlict server.
- Saw window Y update after a few moments (i.e., update in response to a reconnect).
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T4900
Differential Revision: https://secure.phabricator.com/D20656
Summary:
Depends on D20653. Ref T4900. Pass ordering details to the reload endpoint so it can give the client accurate ordering/header information in the response.
The removed comment mentions this, but here's why this is a difficult mess:
- In window A, view a board with "Group by: Owner" and no tasks owned by "Alice". Since "Alice" owns no tasks, this means the columns do not have an "Assigned to: Alice" header!
- In window B, edit task T and assign it to Alice.
- In window A, press "R".
Window A now not only needs to update to properly reflect the state of task T, it actually needs to draw a new "Assigned to: Alice" header in every column.
Fortunately, the "group by" code anticipates this being a big mess, is fairly careful about handling it, and the client can handle this state change and the actual code change here isn't too involved. This is just causing a lot of not-very-obvious indirect effects in the pipeline to handle these situations that need complex redraws.
Test Plan:
- After making various normal edits/creates/moves in window A, pressed "R" in window B. Saw ordering reflected correctly after sync.
- Went through the whole "Group by: Owner" + assign to unrepresented owner flow above. After pressing "R", saw "Assigned to: Alice" appear on the board.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T4900
Differential Revision: https://secure.phabricator.com/D20654
Summary:
Depends on D20652. Ref T4900. When the user presses "R", send a list of cards currently visible on the client and their version numbers.
On the server:
- Compare the client verisons to the server versions so we can skip updates for objects which have not changed. (For now, the client version is always "1" and the server version is always "2", so this doesn't do anything meaningful, and every card is always updated.)
- Compare the client visible set to the server visible set and "remove" any cards which have been removed from the board.
I believe this means that "R" always puts the board into the right state (except for some issues with client orderings not being fully handled yet). It's not tremendously efficient, but we can make versioning better (using the largest object transaction ID) to improve that and loading the page in the first place doesn't take all that long so even sending down the full visible set shouldn't be a huge problem.
Test Plan:
- In window A, removed a card from a board.
- In window B, pressed "R" and saw the removal reflected on the client.
- (Also added cards, edited cards, etc., and didn't catch anything exploding.)
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T4900
Differential Revision: https://secure.phabricator.com/D20653
Summary:
Depends on D20639. Ref T4900. Currently, "BoardResponseEngine" has a `setObjectPHID()` method. This is called after edit operations to mean "we just edited object X, so we know it needs to be updated".
Move toward `setUpdatePHIDs(...)` in all cases, with `setUpdatePHIDs(array(the-object-we-just-edited))` as a special case of that. After this change, callers pass:
- An optional list of PHIDs they know need to be updated on the client. Today, this is always be a card we just edited (on edit/move flows), or a sort of made-up list of PHIDs for the moment (when you press "R"). In the future, the "R" endpoint will do a better job of figuring out a more realistic update set.
- An optional list of PHIDs currently visible on the client. This is used to update ordering details and mark cards for removal. This is currently passed by edit/move, but not by pressing "R" (it will be in the future).
- An optional list of objects. The "R" workflow has to load these anyway, so we can save a couple queries by letting callers pass them. For now, the edit/move flows still rely on the engine to figure out what it needs to load.
This does very little to actually change client behavior, it mostly just paves the way for the next update to the "R" workflow to make it handle add/remove cases properly.
Test Plan:
- Edited and moved cards on a workboard.
- Pressed "R" to reload a workboard.
Neither of these operations seem any worse off than they were before. They still don't fully work:
- When you edit a card and delete the current workboard project from it, it remains visible. This is also the behavior on `master`. This is sort of intentional since we don't necessarily want to make these cards suddenly disappear? Ideally, we would probably have some kind of "tombstone" state where the card can still be edited but can't be dragged, and the next explicit user interaction would clean up old tombstones. This interaction is very rare and I don't think it's particularly important to specialize.
- When a card is removed from the board, "R" can't currently figure out that it should be removed from the client. This is because the client does not yet pass a "visiblePHIDs" state. It will in an upcoming change.
- The "R" flow always sends a full set of card updates, and can not yet detect that some cards have not changed.
- There's a TODO, but some ordering stuff isn't handled yet.
Reviewers: amckinley
Reviewed By: amckinley
Maniphest Tasks: T4900
Differential Revision: https://secure.phabricator.com/D20652
