Summary: I want to reuse this collection view update code in next diff, let's move it to a Helper file to support code sharing.
Reviewed By: calimarkus, iperry90
Differential Revision: D18714061
fbshipit-source-id: a85e8576226276fcab90ea1d2ed23aac5501aa17
Summary:
…uld show empty view
## Changes in this pull request
Basically we can use the existing _itemCountIsZero like all the other usage, and it's faster since it stopped early, compared to before we loop through all the sectionController.numberOfItems and sum up.
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [ ] I added tests, an experiment, or detailed why my change isn't tested.
- [ ] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1400
Differential Revision: D18687637
Pulled By: lorixx
fbshipit-source-id: bcdc57ba43b86cd0ed639ffaab9ae0493fd2f844
Summary:
## Changes in this pull request
`Examples/Examples-iOS/Pods/Pods.xcodeproj/project.pbxproj`
`Examples/Examples-tvOS/Pods/Pods.xcodeproj/project.pbxproj`
this change just Xcode 11.2 generated.
This pr aim to remove useless `IGSystemVersion` file.
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [ ] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1396
Reviewed By: lorixx
Differential Revision: D18638299
Pulled By: iperry90
fbshipit-source-id: 0b6583f1210e5d1570dd672eed8c59901d6785f6
Summary:
As title, let's prepare the 4.0 release with the version bump.
1. Change the Source/Info.plist to have 4.0.0 for CFBundleShortVersionString.
2. Manually update any .md files that used the old release versions.
3. Run `./scripts/setup.sh` to re-update all the pods and the Example projects.
Reviewed By: natestedman
Differential Revision: D18616823
fbshipit-source-id: ca0f2a09cdd9ab592c2cb002fe923e195ce3ab12
Summary:
## Changes in this pull request
Add Swift name annotation `ListAdapterDelegate` to `IGListAdapterDelegate`. The breaking change is written in the `CHANGELOG.md` for the `4.0.0` release
Issue fixed: https://github.com/Instagram/IGListKit/issues/1115
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1116
Reviewed By: bdotdub
Differential Revision: D18278551
Pulled By: iperry90
fbshipit-source-id: db9d1910fc4be7ffa723ee291a9c65db0fd217c0
Summary:
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1377
This splits the diffing-only components from `IGListKit` out into a separate library, `IGListDiffKit`, allowing them to be used independently of the rest of `IGListKit`. We've found that the diffing components of the library are useful independently of the rest of the `UICollectionView` infrastructure, and separating the libraries allows apps to take advantage of those components without paying the full binary size cost of `IGListKit`. The diffing components are available as a subspec in Cocoapods, and as an independent framework in Carthage and direct Xcode project import (i.e. Git submodules).
This is a breaking change, necessitating a major version bump, although it should have only minor impact for projects using the umbrella header `IGListKit.h`, as that header continues to import the diffing utilities. In particular, for a project using Cocoapods and importing `IGListKit.h`, there should be no code changes required.
Reviewed By: calimarkus
Differential Revision: D18114249
fbshipit-source-id: 363b5a427e32800bbc6e82f897230963b4167245
Summary:
Issue: The `NSIndexPath` level updates are really tricky because they don't go through the diffing process. It's up to the caller to make sure that the inserts/deletes/moves all add up to the final number of items. And it's not uncommon that 2 different updates will try to change the same `NSIndexPath` at the same time, which can cause headaches and crashes. For example, if we insert/delete the same `NSIndexPath` twice in the same update pass, we end up with 2 inserts and 1 delete, because we remove duplicated deletes. This causes an `NSInternalInconsistencyException` since things don't add up at the end.
Fix: Lets also remove duplicate `NSIndexPath` inserts to balance things out. In the future, it would be nice to get rid of this API entirely and have everything go through the diffing.
Reviewed By: lorixx
Differential Revision: D17423776
fbshipit-source-id: 62fdf5bd521a915e984e8de180c74858f7ae7ef4
Summary:
* Issue: `[IGListAdapter scrollToObject ...]` currently calls `[collectionView layoutIfNeeded]` before scrolling which creates the current visible cells that will no longer be visible after the scroll. This causes perf issues when we scroll immediately after creating the view controller.
* Fix: Instead of asking the layout object for the attributes, lets go throught the `UICollectionView`. So if the attributes are not ready, the `UICollectionView` will call `-prepareLayout`, return the attributes, but doesn't generate the cells just yet.
Differential Revision: D17326459
fbshipit-source-id: 745942225e0311fea7c3efb07ac1e8b8e0a82996
Summary:
* Issue: Calling `[UICollectionView layoutAttributesForSupplementaryElementOfKind...]` with an indexPath that doesn't have a supplimentary view.
* Cause: `IGListCollectionViewLayout` always returns a non-nil `UICollectionViewLayoutAttributes` when calling `layoutAttributesForSupplementaryViewOfKind` which tells the `UICollectionView` that it's fair game to ask the dataSource for a supplementary view at the indexPath. But when it does, the section controller could return nil, which throws an expection.
In the Apple docs about `[UICollectionViewDataSource collectionView:viewForSupplementaryElementOfKind:atIndexPath:]`
> This method must always return a valid view object. If you do not want a supplementary view in a particular case, your layout object should not create the attributes for that view.
https://developer.apple.com/documentation/uikit/uicollectionviewdatasource/1618037-collectionview?language=objc
* Fix: Just like `UICollectionViewFlowLayout`, if the supplementary view size is empty, lets return a nil attribute.
Reviewed By: Ziewvater
Differential Revision: D17326461
fbshipit-source-id: 507e98f43e951216cf2eafe2449f87df25439e11
Summary:
## Changes in this pull request
Remove the redundant __IPHONE_11_0, because the minimum requirement is Xcode 9.0+ now.
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1351
Differential Revision: D17200815
Pulled By: lorixx
fbshipit-source-id: 25705856cf26d5c215ed956802ab1022c98cbca0
Summary:
### Changes in this pull request
* Improve compilation time by reducing unnecessary compilation dependencies
* More clean code
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1358
Differential Revision: D17200761
Pulled By: lorixx
fbshipit-source-id: 9c6bca4d3d5e2e4b35125c7a12e483d58993c4e8
Summary:
## Changes in this pull request
Update to use `QoS` and take higher priority for responsiveness and energy efficiency.
Following the guideline in the doc [Energy Efficiency Guide for iOS Apps](https://developer.apple.com/library/archive/documentation/Performance/Conceptual/EnergyGuide-iOS/PrioritizeWorkWithQoS.html).
It's better to use `User-initiated` QoS in `Background Diffing` because it focuses on "Work that the user has initiated and requires immediate results".
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1363
Differential Revision: D17200745
Pulled By: lorixx
fbshipit-source-id: 4b9b2159f06edeed92e79bac771fe7b89f6d5708
Summary:
## Changes in this pull request
Delete duplicate headers in `IGListKit.h`
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1347
Reviewed By: joetam
Differential Revision: D16542316
Pulled By: joetam
fbshipit-source-id: feaaaaca0f9eddc4be58996c8a3a2521a7817185
Summary:
## Changes in this pull request
Should map be a map of `IndexPath`s instead of arrays of `IndexPath`s? Also, we might need some unit tests on this part. I don't think it's covered.
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1205
Reviewed By: Ziewvater
Differential Revision: D15962803
Pulled By: lorixx
fbshipit-source-id: f207b69ef0ebad08cd72b14ba53101e56d1604fd
Summary:
- See #1262 (cell reordering using a binding section controller caused problems)
## Changes in this pull request
- Added method override in `IGListBindingSectionController` to update the internal `viewModels` array after a cell has been moved
- Subclasses of IGListBindingSectionController require to call super
Issue fixed: #1262
I've added a test for this change. Is this enough? Or should I add more corner cases or something?
### Checklist
Some tests fail, but they do seem unrelated. I don't know if this is normal. I haven't changed anything regarding those failing tests.
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1274
Reviewed By: candance
Differential Revision: D15592822
Pulled By: lorixx
fbshipit-source-id: d81c18bec0ad0a6a6613089245352bb182a92f26
Summary:
The collection view layout here should not be accessing the data source directly but instead asking the collection view for the number of sections and number of items. This is because the collection view holds an internal cache and if there is a discrepancy between when the collection view cache and the data source, preparing the layout will crash the app because we will send a layout down to the collection view with a items that might not exist in the collection view's cache but exist in the data source.
This diff adds an experiment to see the impact of fixing this check so that when we prepare layouts we don't ask the data source directly but ask the collection view instead. The collection view will then update it's cache and check the data source so it does not crash when we prepare a layout.
```
NSInternalInconsistencyException at __45-[UICollectionViewData validateLayoutInRect:]_block_invoke:
UICollectionView received layout attributes for a cell with an index path that does not exist:
<NSIndexPath: 0xc198a8c3ea391fbb> {length = 2, path = 4 - 0}
```
Reviewed By: lorixx
Differential Revision: D15830610
fbshipit-source-id: f3dae3e87c55b86c3b686309c1144ccfff1375bf
Summary: The test was actually deallocated last year, and this test is never cleaned up. Now let's clean it up.
Reviewed By: chritto
Differential Revision: D15777898
fbshipit-source-id: c3a704ee1bfdd085c4ceb89af424a10ed07a65b6
Summary:
## Changes in this pull request
Issue fixed: https://github.com/Instagram/IGListKit/issues/1275
This PR is straintforward solution as mentioned in https://github.com/Instagram/IGListKit/issues/1275. When I finished it, I realized that it is not a problem in `IGListCollectionViewLayout Partial Optimization` and `IGListBindingSectionController` deserved it. Maybe `IGListCollectionViewLayout` is rarely used because of great builtin UICollectionViewFlowLayout or cells in section rarely need changing their sizes in practices.
Despite it is not `IGListCollectionViewLayout`'s fault, I think it can be more optimized against `invalidateLayoutWithContext`. I would like to make an another PR to optimize it.
Due to this PR just giving a proposed solution, it lacks of adding tests and changing log. When this solution is accepted, I would like to complete these todos.
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1285
Reviewed By: candance
Differential Revision: D15592807
Pulled By: lorixx
fbshipit-source-id: ae06abce896341509de4f3dfb73b3a7bc0a68c51
Summary:
The implementation of
```Objective-C
-[IGListAdapter scrollToObject:supplementaryKinds:scrollDirection:scrollPosition:animated:]
```
has little inconsistencies where the `UICollectionViewScrollPositionLeft` and `UICollectionViewScrollPositionTop` considering the content inset left/top of the collection view was being applied to the final offset. However, the `UICollectionViewScrollPositionRight` and `UICollectionViewScrollPositionBottom` ignoring the content inset right/bottom of the collection view was being applied to the final offset. The different result is that scrolling to the most `Left/Top`, the first section content always be visible by considering its content inset while scrolling to the most `Right/Bottom`, the last section content always be fully displayed but not be fully visible by considering its content inset.
## Changes in this pull request
Issue fixed: #
### Checklist
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1284
Reviewed By: lorixx
Differential Revision: D13590416
Pulled By: lorixx
fbshipit-source-id: ae52be9e5ba25b50c7a0ad768a4af728347523e2
Summary:
This is a bigger PR than I'd like but there is a bunch of stuff in here that feels like we should land all at once to fix the Travis setup.
Currently, there is a few things wrong. Briefly:
1. Travis is failing to even build the library (see: [build 2298](https://travis-ci.org/Instagram/IGListKit/jobs/506564900))
2. There are legitimately failing tests
3. Travis seems to be flakey running UI tests
This PR deals with the first two.
The Xcode projects were not building because some recently added files weren't added to the xcodeproj (ie. 46a124ddfe and 4662454c4a).
Also added `Foundation.h` import for `bool` definition for `IGSystemVersion` (also a new file)
The main `Podfile.lock` was already being generated via `1.5.3` (0c0b31ad7c) but the `Gemfile` had the wrong version and none of the example project `Podfile.lock`s were re-generated with it.
Once I got the projects correctly building and tests running, I discovered there were a bunch of tests that were legitimately failing. It's unclear when these started failing and what caused it.
I thought it made sense to disable these tests for now and then create Github issues to fix them as starter tasks instead of making this diff even larger.. I can create those if that sounds good.
Somewhat related, it seems like the UI tests on Travis may be a bit flakey. They pass locally, but breaks on Travis [[example](https://travis-ci.com/bdotdub/IGListKit/jobs/195758250)] with error messages like `Failure requesting automation session for com.instagram.IGListKitExamples:85190`
----
- [X] All tests pass. Demo project builds and runs.
- [X] I added tests, an experiment, or detailed why my change isn't tested.
- [X] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [X] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1317
Reviewed By: rnystrom
Differential Revision: D15157375
Pulled By: rnystrom
fbshipit-source-id: b29131f59b74398b6d2b3a73453248cb3325a955
Summary:
* Currently, there's no easy way to measure how long we spend dequeueing cells or handling scroll events across the app. This makes it a bit difficult to catch scroll-perf issues early on.
* Fix: Lets create `IGListAdapterPerformanceDelegate` which gives us that information.
Reviewed By: rnystrom
Differential Revision: D14430449
fbshipit-source-id: 7bc170bdaa37807c8719976e8818f3a578ee17dd
Summary:
We have observed that keeping the coalescing time here would actually decrease the stability here, it makes the app crash more often.
It's probably caused by some race condition during that time and UICV updates become out of sync.
Reviewed By: rnystrom
Differential Revision: D14379265
fbshipit-source-id: 502c1c14fb8bdfc35969f721687e82888b160110
Summary:
This assert was hoping to find if there are any duplicates from the `fromObjects`. However, from the finding, we did not see any assertion or trace that this is the case.
The `objectsWithDuplicateIdentifiersRemoved` is actually doing extra work during updates.
We should remove this for the sake of better scroll perf and faster UI update.
Reviewed By: calimarkus
Differential Revision: D14339124
fbshipit-source-id: ff40fd23a05640058673d46f63a9e4bf516dd3f6
Summary: This change allows arbitrary layouts to be used in tandem with IGListCollectionView and recieve udpated section indices, by conforming to to conform IGListCollectionViewLayoutCompatible.
Reviewed By: maxoll
Differential Revision: D13671204
fbshipit-source-id: e030f413aefff71754731d1259037077ffb73959
Summary:
Issue fixed: #1288
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1289
Reviewed By: timonus
Differential Revision: D13590406
Pulled By: rnystrom
fbshipit-source-id: 0814b0c2383a185da747176bad57dfb15c7e883f
Summary: This diff adds a new `IGListExperimentBackgroundDiffingSerial` experiment that behaves similarly to `IGListExperimentBackgroundDiffing` but executes changes in a per-`IGListAdapterUpdater` serial queue instead of a global queue. Global queues are concurrent, and concurrent queues by nature don't ensure the execution order of the blocks that are enqueued into them, so there's a chance that the former `IGListExperimentBackgroundDiffing` experiment would lead to changes being applied out of order.
Reviewed By: calimarkus
Differential Revision: D13567403
fbshipit-source-id: a2aebb20e5e7dc22601fad3c2dfc80112747c81b
Summary:
Some calls to cellForItemAtIndex: on IGListSectionController expect this method to return a reference to an existing, already dequeued cell. It doesn't do this - it dequeues a new cell instance and returns that.
Added a warning note to the docs.
Reviewed By: lorixx
Differential Revision: D13222395
fbshipit-source-id: bb7a80f9f005e2675eef3d01a236a5f9f0008b46
Summary:
This happens when we set collectionView.dataSource to be nil before calling performUpdates:, the crash is easily reproduced by a simple unit test.
To make the infra more robust and we don't crash the app, let's add an early return before applying a collectionView update.
Reviewed By: rnystrom, calimarkus
Differential Revision: D12896196
fbshipit-source-id: ab024d0e7e9282d50c3be297e1e67cfccaff8242
Summary:
Issue fixed: #1235
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1236
Reviewed By: lorixx
Differential Revision: D12839081
Pulled By: rnystrom
fbshipit-source-id: 3e9e1192f16912d560d76f3730a377f303708cd7
Summary:
Issue fixed: #1219
No tests were added and no entry was added to `CHANGELOG.md` since this is a trivial change.
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1242
Differential Revision: D12840487
Pulled By: rnystrom
fbshipit-source-id: ded4b09765c5666e23a9aa3596328e90df2b3229
Summary:
Issue fixed: #1223
- [x] All tests pass. Demo project builds and runs.
- [x] I added tests, an experiment, or detailed why my change isn't tested.
- [x] I added an entry to the `CHANGELOG.md` for any breaking changes, enhancements, or bug fixes.
- [x] I have reviewed the [contributing guide](https://github.com/Instagram/IGListKit/blob/master/.github/CONTRIBUTING.md)
Pull Request resolved: https://github.com/Instagram/IGListKit/pull/1241
Differential Revision: D12838898
Pulled By: rnystrom
fbshipit-source-id: af47550dd320fb0e813e8ae22bc7051c2460a05d
Summary: This would help us understands what was the diff result from the algorithm and compare it with the `updates` to check what went wrong. Essentially, this would make our debugging easier going forward.
Reviewed By: calimarkus
Differential Revision: D12828892
fbshipit-source-id: 6dc52cdba1adb5a841760b51599e2df4c845c364
Summary:
There are some crashes related to this path when the number of sections are changed when using the `preferItemReloadsForSectionReloads` optimization for item reloads.
For now, this change would just limit the flag only useable when the section count is unchanged. The proper fix would be finding a repro-case for the unsafe trace.
The follow-up here is adding more logging to capture the crash reason and investigate any variable that leads to this crash.
Reviewed By: calimarkus
Differential Revision: D12828061
fbshipit-source-id: 040e66c051c54f889ee9fbc3e4a48ab6bf93e162
Summary: Provide additional context in willPerformBatchUpdates, which can be useful for debugging.
Reviewed By: lorixx
Differential Revision: D10436588
fbshipit-source-id: a59803affc24ca20f7726e36117df6126b984afb
Summary: Verify our assumption that the `fromObjects` should never contain duplicate identifiers.
Reviewed By: calimarkus
Differential Revision: D10410190
fbshipit-source-id: 4952698fee6373224b4c8395778009540c5ae380
Summary:
section moves are extremely unsafe and crashy.
If we still use the `preferItemReloadsForSectionReloads` on, the item reloads won't be correct when we applies the same logic for cases without section moves.
It leads to crash:
```
2018-09-04 23:28:12.184875-0700 xctest[51440:12103918] *** Terminating app due to uncaught exception 'NSInternalInconsistencyException', reason: 'Invalid update: invalid number of items in section 0. The number of items contained in an existing section after the update (1) must be equal to the number of items contained in that section before the update (1), plus or minus the number of items inserted or deleted from that section (0 inserted, 1 deleted) and plus or minus the number of items moved into or out of that section (0 moved in, 0 moved out).'
```
even if we change to use -reloadSections: instead, we still get crash like the following:
```
2018-09-04 23:21:14.129961-0700 Direct[45654:12078189] *** Terminating app due to uncaught exception 'NSInternalInconsistencyException', reason: 'attempt to perform a delete and a move from the same section (25)'
```
Thus let's only add the moves.count == 0 check before using the prefer item reloads over section reload optimization.
Reviewed By: calimarkus
Differential Revision: D9661723
fbshipit-source-id: 8c3ddcd5178cb1aa58ccd9ae274cd8e5198f7609
Summary:
This made the items reload safer, as I tested that passing in an out-of-bound section index to -[UICollectionView numberOfItemsInSection:] would crash the app.
UIKit did not handle this gracefully.
Reviewed By: calimarkus
Differential Revision: D9624061
fbshipit-source-id: bcf14a3d4746b648046992adf221faff36a0257c
Summary:
When we added the ability to register the same class with multiple reuse identifiers, forgot to adjust the `registeredClasses` set to reflect that. This resulted in a bug where the collection view could be asked for a cell of a class that had already been registered, but with a different reuse identifier, which would end in a crash.
This diff fixes the crash, I'll follow up with another diff that adds unit tests around this, but want to fix master ASAP.
Differential Revision: D9560338
fbshipit-source-id: e8df6fe5d9929b44842c6abc01d52064ccb4147e
