I audited this more carefully and found some gaps in my PR from yesterday.
This also ensures the grouping and ordering of all the contents properties
is consistent across all of the places they appear.
In the future, it would be great to make the #define macros more sophisticated
so that we can add property bridging with fewer lines changed, and for
objects besides ASDisplayNode too!
This makes it much easier to use the ASCoreAnimationExtras method which offers by
far the most efficient way to display a stretchable image.
In the future, we should move that function to UIImage+ASConvenience or another
header where it can be more easily found and enjoyed!
Texture fails to build with this error on clang 4.0:
```
external/Texture/pod_support/Headers/Public/AsyncDisplayKit/ASDispatch.h:32:35: error: illegal initializer type 'atomic_size_t' (aka '_Atomic(size_t)')
__block atomic_size_t counter = ATOMIC_VAR_INIT(0);
^
In module 'std' imported from external/Texture/pod_support/Headers/Public/AsyncDisplayKit/ASStackUnpositionedLayout.h:18:
/private/var/tmp/_bazel_bkase/a00d4cbe29902fb63d5778cc19944cd2/external/clang40/bin/../include/c++/v1/atomic:1839:30: note: expanded from macro 'ATOMIC_VAR_INIT'
^
1 error generated.
```
See
http://techqa.info/programming/question/38233019/Initializing-an--atomic-int--with-a-braced-constant--Is-this-valid-C-code--If-so-why-does-it-not-compile-in-clang-
Replacing the intialization with just 0 (and not the macro fixes the
error). For now this is safe because the macro just expands to the
value.
* Minor refactors in ASCollectionView and its private cell classes
- `_ASCollectionReusableView` and `_ASCollectionViewCell` no longer expose getters for their collection element and cell node properties. This is to make sure that clients can only obtain elements from the visible map which is always the source of truth.
- Since the map can return `nil` for an element request, it's much safer to check and avoid adding/removing a nil pointer to an `NSArray`.
- Since we use a special way to check whether an object of kind of `_ASCollectionViewCell` or `_ASCollectionReusableView`, based on the assumption that these classes are subclass restricted, I added cast-or-return macros in the header files, closer to where the restrictions are declared.
* Add ASDynamicCastStrict
* Add element and node properties back to _ASCollectionReusableView and _ASCollectionViewCell
* Assert unexpected nil elements
* Always mark an element visible even if it is backed by an UIKit / non-_ASCollection* view
* Fix typo
* Remove unnecessary changes
* Dump mistakes
* Update CHANGELOG
* Can't track visibility of elements backed by non-_AS views
Subclasses of NetworkAddress are deleted using pointers to the base class,
so a virtual destructor is required. Also fix a couple of warnings on 64bit.
This was originally added for ASCollectionNode and ASTableNode preloading to work
as intended when nested inside of another ASRangeController-powered node. Indeed,
it is still necessary to trigger layout on these UIKit-backed components in order
for their own ASRangeControllers to start preparing content within them.
However, unlike the comment suggests, it is *not* necessary to do this for image
nodes. ASNetworkImageNode has only one .URL, and does not use the .bounds or
.calculatedLayout at all during loading. Even the ASMultiplexImageNode does not
use the .bounds, and the ASMapNode uses .calculatedLayout instead of .bounds.
This change has important performance benefits. In particular, it avoids
layouts that would occur on the main thread, often including text sizing,
and also can result in redundant layout passes (even during a layout pass that
triggers a node to enter the preload range, it may force its own layout early).
It would be great to test this change with Pinterest to confirm its safety, but
based on a full audit of the framework codebase, the only possibility that I
see for a regression is if app implementations of -didEnterPreloadState make
direct use of .bounds instead of .calculatedLayout (which is easy to fix).
* Fix a crash in table view caused by executing an empty change set during layoutSubviews
- Previously, when a change set is empty, `ASDataController` forwards the change set to its delegate right away, without dispatching to its editing queue and then back to main.
- This behaviour can potentially cause bad internal states in UITableView which trigger a crash reported in https://github.com/TextureGroup/Texture/issues/83.
- Fix by still reusing the existing pending map, because the data source's state has not changed, but go through the editing queue and main queue tunnel.
* Update CHANGELOG
I believe this check is supposed to be here. Unit tests pass with it in place.
Without it, calling layoutThatFits: (as ASDataController does) and then calling
it again later (as ASCollectionNode does when answering the sizeForItem: call)
will recompute the layout, potentially on main.
This seems to have more of an impact / benefit for Yoga layouts, but I don't think
its benefit is exclusive to them.
* [ASTextNode2] Provide compiler flag to enable ASTextNode2 for all usages.
The runtime switch is helpful, but is too slow to be shipped in a large
production application where startup time is carefully optimized.
Although this doesn't pass text-related snapshot tests when enabled, it
does allow apps to rely on built-in components like ASButtonNode using
the same text stack as when they are manually creating ASTextNode2
instances.
A simpler approach would be to use a #define ASTextNode ASTextNode2,
but this would create unusual keyword highlighting in code referencing
ASTextNode. However, because it would be less invasive and this is
not on by default, we could do that instead if preferred.
* Update AsyncDisplayKit.h
* [ASTextNode2] Improve naming and documentation of ASTEXTNODE_EXPERIMENT_GLOBAL_ENABLE
* [ASTextNode2] CHANGELOG.md for #410.
* [ASTextNode2] Add initial implementation for link handling.
This is a fairly basic first step to achieving feature parity between ASTextNode2
and ASTextNode. It does not yet do the 9-box detection of links, and there is
other code from ASTextNode that could be shared to improve this.
However, in the interest of getting a shippable implementation running as soon
as possible, I'm hoping to quickly refine this. Then we can continue improving
it based on the original ASTextNode with the benefit of testing UI against it.
* [ASTextNode2] Refine implementation of link handling.
This ensures that measure funcs are not set for container / empty spacer
nodes, because Yoga has more internal capabilities than layoutThatFits:
knows about.
Avoid need for the main layout pass to directly call setup for measure
funcs, by making it an implicit part of .yogaLayoutInProgress =.
Tear down the measure func after the layout pass to avoid retain cycle.
