Summary:
Fixes#10433
The code didn't account for the fact that cancelLoad is set by two different threads. It gets set on the URL request queue when the request completes successfully and it gets set on the UI queue during cancelation. This oversight lead to a couple of different kinds of crashes.
1. Attempt to invoke a nil function -- We check that cancelLoad is non-nil and on the next line we call it. However, cancelLoad could have been set to nil by the other thread between the time it was verified to be non-nil and the time it was called. Consequently, the program will attempt to call nil and crash.
2. Block deallocated while it's executing -- Suppose cancelLoad points to a block, it is verified to be non-nil, and it is successfully invoked. In the middle of executing the block, cancelLoad, the last reference to the block, is set to nil. This causes the block to be deallocated and its captured values to be freed. However, the block continues executing and the next time it attempts to use a captured value
Closes https://github.com/facebook/react-native/pull/11145
Differential Revision: D4261499
Pulled By: javache
fbshipit-source-id: 46424c6dd8cfa085cef32d945308de07798040bc
Summary:
To make React Native play nicely with our internal build infrastructure we need to properly namespace all of our header includes.
Where previously you could do `#import "RCTBridge.h"`, you must now write this as `#import <React/RCTBridge.h>`. If your xcode project still has a custom header include path, both variants will likely continue to work, but for new projects, we're defaulting the header include path to `$(BUILT_PRODUCTS_DIR)/usr/local/include`, where the React and CSSLayout targets will copy a subset of headers too. To make Xcode copy headers phase work properly, you may need to add React as an explicit dependency to your app's scheme and disable "parallelize build".
Reviewed By: mmmulani
Differential Revision: D4213120
fbshipit-source-id: 84a32a4b250c27699e6795f43584f13d594a9a82
Summary:
This PR is related to the multitude of crashes (#10016, #9751, #9882).
From my understanding, we should be using a strong reference when calling `decodeImageData` or we could be calling the method on a deallocated instance.
PR #9751 have mitigated this by adding a fail-safe, but I think the culprint is the weak reference, which this PR fixes.
Tested on iOS only, since it doesn't touch Android.
Closes https://github.com/facebook/react-native/pull/10147
Differential Revision: D3938763
fbshipit-source-id: 7389d4ae7a98926014401a1fe0cbbdcdd5ee6a01
Summary:
This fixes https://github.com/facebook/react-native/issues/9751
We were seeing the same issues as the original reporter.
I'm not an expert on this code, so apologies if it's naive, but we haven't seen the crash since making this change.
From my understanding of it, it seems like the `cancelLoad` block was being released before the block returned by `_loadImageOrDataWithURLRequest:` was called.
This PR checks the `cancelled` flag before calling `cancelLoad()`.
Closes https://github.com/facebook/react-native/pull/10016
Differential Revision: D3902723
Pulled By: javache
fbshipit-source-id: 75cd115e28694105c6fc29469986998ca0d4cd09
Summary:
There is a conflict between RCTImageLoader and RCTAssetsLibraryRequestHandler on handling images, which was caused by making RCTPhotoLibraryImageLoader able to handle assets library requests as well. This gives more priority to RCTImageLoader instead.
Fixes#9031.
Reviewed By: mmmulani
Differential Revision: D3627451
fbshipit-source-id: 7ffd2c66f43ce1479c9a117768fb2d29f9d0dc08
Summary:
Before this PR, ```RCTImageLodaer```'s Cache was too big(200MB on disk) and It doesn't work with HTTP Cache-Control header. So to provide dynamic image, the users must have to add random value on url( ex. adding current date) to avoid cache.
So I change that cache system to default ```NSURLRequest```'s cache system, which is well-working with HTTP specs. As the discussion on this issue #7571 , making custom cache policy processor is not ready yet and useless, over-tech things, I think.
Even we have no plan about image cache system(or would change plan later), before having a nice plan, I think we should let user use image module with common HTTP Specs.
So I remove custom ```NSURLCache```, and make logic like below,
1. try fetch image,
2. on response, get ```Date``` on response's header and make ```cacheKey``` with ```Date```.
> (why? because if ```NSURLRequest```'s response was cached, the response's ```Date``` header dosen't change.)
3. find decoded imag
Closes https://github.com/facebook/react-native/pull/8235
Reviewed By: bnham
Differential Revision: D3469086
Pulled By: javache
fbshipit-source-id: 35a5552cda6e6c367481020bbf3c28eb4a9d0207
Summary:
This uses `[UIImage imageNamed:]` to load local assets that are bundled using `require('../image/path.png')` and makes sure it is done synchronously on the main queue to prevent images from flickering. This improves user experience a lot when using large local images and prevents icon flickers to match the behaviour of most native apps.
This adds to methods to the ImageLoader protocol, one to tell if the image loader must be executed on the url cache queue and one to tell if the result of the image loader should be cached. I then use these to make the LocalImageLoader bypass the url cache queue and avoid caching images twice.
Note that this doesn't affect debug builds since images are loaded from the packager.
I'm not sure if we want to still support async loading of local images as I'm not sure how much of a perf difference this will make. Maybe someone at fb can benchmark this see how it affects your apps but there wasn't a noticeable one in mine. Also I only enabled this for loading png and jpg im
Closes https://github.com/facebook/react-native/pull/8102
Reviewed By: bnham
Differential Revision: D3433647
Pulled By: javache
fbshipit-source-id: 37bd6aff20c0465c163db3cdbcaeaedff55f7b1f
Summary:
Under rare and as-yet-to-be determined circumstances, images can sometimes fail to load/download and get "stuck", without producing an error.
Because the `RCTNetworkTask` for these images is stuck in the "in progress" state, they clog up the RCTImageLoader task queue, which has a limit of 4 concurrent in-progress tasks.
This was previously masked by the fact that we automatically cancelled image requests when the RCTImageView moved offscreen, but we no longer do that.
This diff adds logic to detect some types of stuck task and remove them, thereby unblocking the queue. I've also restored the functionality of cancelling downloads for offscreen images (but not unloading the image itself) so that stuck images will be cancelled when you move to another screen, instead of using up space in the queue forever.
Reviewed By: fkgozali
Differential Revision: D3398105
fbshipit-source-id: 75ee40d06a872ae8e1cb57f02f9cad57c459143c
Summary:
Now that we no longer have a separate NetworkImageView implementation, we can remove that code path from Image.js
I've also moved the prefetch method into RCTImageViewManager for consistency with the getImageSize method, which means we no longer need to export the RCTImageLoader module to js.
Reviewed By: javache
Differential Revision: D3398157
fbshipit-source-id: fbbcf90a61549831ad28bad0cb3b50c375aae32c
Summary:
As per https://twitter.com/olebegemann/status/738656134731599872, our use of "main thread" to mean "main queue" seems to be unsafe.
This diff replaces the `NSThread.isMainQueue` checks with dispatch_get_specific(), which is the recommended approach.
I've also replaced all use of "MainThread" terminology with "MainQueue", and taken the opportunity to deprecate the "sync" param of `RCTExecuteOnMainThread()`, which, while we do still use it in a few places, is incredibly unsafe and shouldn't be encouraged.
Reviewed By: javache
Differential Revision: D3384910
fbshipit-source-id: ea7c216013372267b82eb25a38db5eb4cd46a089
Summary:
Allows developers to specify headers to include in the HTTP request
when fetching a remote image. For example, one might leverage this
when fetching an image from an endpoint that requires authentication:
```
<Image
style={styles.logo}
source={{
uri: 'http://facebook.github.io/react/img/logo_og.png',
headers: {
Authorization: 'someAuthToken'
}
}}
/>
```
Note that the header values must be strings.
Works on iOS and Android.
**Test plan (required)**
- Ran a small example like the one above on iOS and Android and ensured the headers were sent to the server.
- Ran a small example to ensure that \<Image\> components without headers still work.
- Currently using this code in our app.
Adam Comella
Microsoft Corp.
Closes https://github.com/facebook/react-native/pull/7338
Reviewed By: javache
Differential Revision: D3371458
Pulled By: nicklockwood
fbshipit-source-id: cdb24fe2572c3ae3ba82c86ad383af6d85157e20
Summary: The decoded image cache was previously static, meaning that cached images could persist beyond the lifetime of the module. This resulted in some flakiness in the RCTImageLoaderTests due to the loader returning cached image instanced from previous tests instead of the correct instance.
Reviewed By: bestander
Differential Revision: D3346329
fbshipit-source-id: 375af8894cef1c5b6303c6cdfd7eb57ebcfe3251
Summary:
In RN we cache image data after loading/downloading an image, however the data we store is the compressed image data, and we decode this asynchronously each time it is displayed.
This can lead to a slight flicker when reloading image components because the decoded image is discarded and then re-decoded.
This diff adds a small (5MB) cache for decoded images so that images that are currently on screen shouldn't flicker any more if the component is reloaded.
Reviewed By: bnham
Differential Revision: D3305161
fbshipit-source-id: 9969012f576784dd6f37d9386cbced2df00c3e07
Summary: Fixed a crash where making a network request with unrecognized/unsupported scheme would crash by attempting to insert a nil RCTNetworkTask into an array.
Reviewed By: javache
Differential Revision: D3275810
fb-gh-sync-id: be208c6bf87d882209901b850dcc25da4ba79a08
fbshipit-source-id: be208c6bf87d882209901b850dcc25da4ba79a08
Summary:Adds `Image.prefetch` to prefetch remote images before they are used in an actual `Image` component. This is based off of #4420 by sospartan and skevy's work.
Closes https://github.com/facebook/react-native/pull/6774
Differential Revision: D3153729
Pulled By: bestander
fb-gh-sync-id: ef61412e051a49b42ae885edce7905a8ca0da23f
fbshipit-source-id: ef61412e051a49b42ae885edce7905a8ca0da23f
Summary:Initializing native modules can block the main thread for tens of milliseconds when it starts up, making it difficult to instantiate the bridge on demand without causing a performance blip.
This diff splits up the initialization of modules so that - although they still happen on the main thread - they don't block the thread continuously.
Reviewed By: javache
Differential Revision: D2965438
fb-gh-sync-id: 38c9c9d281e4672b5874d68b57d4c60d1d268344
shipit-source-id: 38c9c9d281e4672b5874d68b57d4c60d1d268344
Summary:
public
Standardises the image decoding logic for all image sources, meaning we get the benefits of efficient downscaling of images from all sources, not just ALAssets.
Reviewed By: javache
Differential Revision: D2647083
fb-gh-sync-id: e41456f838e4c6ab709b1c1523f651a86ff6e623
Summary:
public
This diff adds a `getSize()` method to `Image` to retrieve the width and height of an image prior to displaying it. This is useful when working with images from uncontrolled sources, and has been a much-requested feature.
In order to retrieve the image dimensions, the image may first need to be loaded or downloaded, after which it will be cached. This means that in principle you could use this method to preload images, however it is not optimized for that purpose, and may in future be implemented in a way that does not fully load/download the image data.
A fully supported way to preload images will be provided in a future diff.
The API (separate success and failure callbacks) is far from ideal, but until we agree on a unified standard, this was the most conventional way I could think of to implement it. If it returned a promise or something similar, it would be unique among all such APIS in the framework.
Please note that this has been a long time coming, in part due to much bikeshedding about what the API should look like, so while it's not unlikely that the API may change in future, I think having *some* way to do this is better than waiting until we can define the "perfect" way.
Reviewed By: vjeux
Differential Revision: D2797365
fb-gh-sync-id: 11eb1b8547773b1f8be0bc55ddf6dfedebf7fc0a
Summary: public
The `bridge.modules` dictionary provides access to all native modules, but this API requires that every module is initialized in advance so that any module can be accessed.
This diff introduces a better API that will allow modules to be initialized lazily as they are needed, and deprecates `bridge.modules` (modules that use it will still work, but should be rewritten to use `bridge.moduleClasses` or `-[bridge moduleForName/Class:` instead.
The rules are now as follows:
* Any module that overrides `init` or `setBridge:` will be initialized on the main thread when the bridge is created
* Any module that implements `constantsToExport:` will be initialized later when the config is exported (the module itself will be initialized on a background queue, but `constantsToExport:` will still be called on the main thread.
* All other modules will be initialized lazily when a method is first called on them.
These rules may seem slightly arcane, but they have the advantage of not violating any assumptions that may have been made by existing code - any module written under the original assumption that it would be initialized synchronously on the main thread when the bridge is created should still function exactly the same, but modules that avoid overriding `init` or `setBridge:` will now be loaded lazily.
I've rewritten most of the standard modules to take advantage of this new lazy loading, with the following results:
Out of the 65 modules included in UIExplorer:
* 16 are initialized on the main thread when the bridge is created
* A further 8 are initialized when the config is exported to JS
* The remaining 41 will be initialized lazily on-demand
Reviewed By: jspahrsummers
Differential Revision: D2677695
fb-gh-sync-id: 507ae7e9fd6b563e89292c7371767c978e928f33
Summary: public
A missing return statement in RCTImageLoader meant that cached images would be loaded twice - once from cache and again from the source.
This was mostly innocuous, causing only a slight perf regression due to the image cache being effectively disabled, however in some cases (such as RCTImageEditingManager.cropImage) it caused the success callback to fire twice, resulting in a crash.
Reviewed By: fkgozali
Differential Revision: D2684956
fb-gh-sync-id: 7580a6fbfe00a30807951803e04bfcdbee3bb80a
Summary: public
I had previously assumed (based on past experience and common wisdom) that `[UIImage imageWithData:]` was safe to call concurrently and/or off the main thread, but it seems that may not be the case (see https://github.com/AFNetworking/AFNetworking/pull/2815).
This diff replaces `[UIImage imageWithData:]` with ImageIO-based decoding wherever possible, and ensures that it is called on the main thread wherever that's not possible/convenient.
I've also serialized access to the `NSURLCache` inside `RCTImageLoader`, which was causing a separate-but-similar crash when loading images.
Reviewed By: fkgozali
Differential Revision: D2678369
fb-gh-sync-id: 74d033dafcf6c412556e4c96f5ac5d3432298b18
Summary: public
Removed redundant calls to [RCTNetwork canHandleRequest] in release mode when loading images, and improved perf for handler lookups when running in debug mode.
Reviewed By: tadeuzagallo
Differential Revision: D2663307
fb-gh-sync-id: 13285154c1c3773b32dba7894d86d14992e2fd7d
Summary: public
Added lightweight genarics annotations to make the code more readable and help the compiler catch bugs.
Fixed some type bugs and improved bridge validation in a few places.
Reviewed By: javache
Differential Revision: D2600189
fb-gh-sync-id: f81e22f2cdc107bf8d0b15deec6d5b83aacc5b56
Summary: public
The image loader was previously returning on the main thread, which could lead to poor performance due to various call sites doing further image processing (resizing, cropping, etc.) directly in the completion block.
This diff modifies the loader to return on a background thread (the same one used to load the image), and updates the call sites to dispatch to the explicit thread they need.
Reviewed By: javache
Differential Revision: D2549774
fb-gh-sync-id: fed73b7c163fdf67ff65bae72ab1986327e75815
Summary: public
Added RCTDataRequestHandler, which is responsible for loading data URLs. This moves the logic for data URL handling out of RCTImageDownloader (no longer needed) and into the RCTNetwork library, where it makes more sense.
This also means that it is now possible to load data URLs via XHR, and use them for purposes other than just images.
Reviewed By: javache
Differential Revision: D2540964
fb-gh-sync-id: 4f0418bd6b9186f047cc8297276bb970795af104
Summary: @public
This diff unifies the logic for detecting when images refer to XCAsset files into a single function (RCTXCAssetNameForURL) and uses it for both +[RCTConvert UIImage:] and RCTImageLoader.
I've also tightened the definition of XCAssets so that it only applies to images inside .car files, not any image inside the main bundle. This avoids using the +[UIImage imageNamed:] when not strictly necessary, which is desirable since that method is not thread-safe, and has undocumented caching behavior that is difficult to reason about.
Reviewed By: @javache
Differential Revision: D2526400
fb-gh-sync-id: 7199c2a44f1d55ff236d2c38a0a9368739b993d5
Summary: @public
This diff implements inline image support for <Text> nodes. Images are specified using <Image> tags, however all properties of the image are currently ignored apart from the source (including width/height styles).
Images are loaded asyncronously, and will trigger a text re-layout when they have loaded.
Reviewed By: @javache
Differential Revision: D2507725
fb-gh-sync-id: 59d0696d00a1bc531915cc35242a16b2dec96e85
Summary:
GIF images are currently loaded as a CAKeyframeAnimation, however returning this animation directly from RCTImageLoader was dangerous, as any code that expected a UIImage would crash.
This diff changes RCTGIFImageLoader to return a UIImage of the first frame, with the keyframe animation attached as an associated object. This way, code that is not expecting an animation will still work correctly.
Adam Comella
Pieter De Baets
Alex Kotliarskyi
Mattias Pfeiffer
Toby Cox
Mehdi Mulani
Maria Mateescu
Nathan Azaria
Scott Wolchok
skatpgusskat
Janic Duplessis
yeshuang
Skotch Vail
Nick Lockwood
Maarten Schumacher
Adam Comella
Howard Yeh
James Ide
Marc Horowitz
Dral
Alex Akers
