Summary:
Since we are reading from a file, we should make sure this struct is packed, just in case we change it down the line and the compiler decides it might want to introduce padding, we're now protected against that.
There was also a discussion about the fact that people might use `ptr += sizeof(BundleHeader)` as an idiom in their code, which would currently be incorrect, if padding was introduced at the end of the file. Actually, it remains incorrect to do that now, because a RAM bundle header is a different size to a BC Bundle header. If people are properly testing their code, they should spot this pretty quickly, because it will always be an incorrect thing to do with a RAM bundle, so this isn't as bad as previously thought: where the code only succeeds when the compiler deigns to not pad the struct at the end.
This diff also cleans up how headers are initialised. `BundleHeader` has a constructor that explicitly zero-initialises it so we can rely on the default initializer to do the right thing now.
Reviewed By: mhorowitz
Differential Revision: D4572032
fbshipit-source-id: 7dc50cfa9438dfdfb9f842dc39d8f15334813c63
Summary: I broke Circle CI builds by moving a file, so I am updating the OSS build setup with details of the move. Whilst I was testing the change, I noticed that the UI Explorer build was also already broken, so this diff also changes some of its configuration to make it build again.
Reviewed By: javache
Differential Revision: D4579137
fbshipit-source-id: 4cbb1ef148075280b991cbc5bb47bf96ec3d543a
Summary:
In order for `CxxModule` modules to be able to do things like emit events, we need to expose the bridge instance.
To allow classes derived from `CxxModule` in other projects to include `<cxxreact/Instance.h> I also needed to convert all the header includes to non-relative ones.
Reviewed By: javache
Differential Revision: D4564145
fbshipit-source-id: a5bc28dd9b40e2b141af9e867105c56083fbdcdc
Summary:
`JSIndexedRAMBundle` used to read scripts including the terminating null bytes. That worked on iOS, but not on Android, where we have an ascii-only optimization.
This changes that behavior to only read the actual script data excluding the terminal null.
Reviewed By: javache
Differential Revision: D4265374
fbshipit-source-id: 7e6f943666aee610d79939cef09b103305803c69
Summary:
The C++ standard requires that when a function is used in a template it's prototype needs to be defined not only before the template specialization, but also before the template itself.
Because of that one needs to (in certain compilers) be aware of the proper order of includes so that the function prototype is defined before the JSCExecutor.h is included.
As a workaround the toValue might be written as a template (ValueEncoder<T>::toValue) defined in JSCExecutor.h instead of being an non-existing symbol.
Thanks to that the JSCExecutor.h does not have to be included before the specialization of the ValueEncoder template.
Reviewed By: mhorowitz
Differential Revision: D4182724
fbshipit-source-id: 9bdf239ae66ef7a7d2c82daf7db5926472687bde
Summary:
`JSBigMmapString` needs to know too much about the details of the "optimized bundle" format. In honour of this fact, I'm renaming it to `JSBigOptimizedBundleString`. In a following diff, I will introduce a new class just for File-backed Strings, whose only responsibility is dealing with an mmaped region.
This diff already pulls in some fixes for formatting reference and pointer declarations from the linter.
Reviewed By: michalgr
Differential Revision: D4189391
fbshipit-source-id: b376c2a8d9ae5b83575da8457e607bbbfc648ebd
Summary:
Introduces the inspector library supporting the Chrome Debugging Protocol for JavaScriptCore. Eventually this will mean that it is possible to attach
the Chrome inspector directly to the JSC instance running on the device. This library doesn't define the actual transport but leaves that up to the platform
layer.
The main entry point (and the only exported header) is `Inspector.h`.
This diff only introduces the basics supporting the `Schema` and `Inspector` domains meaning it doesn't have any features yet. These will come in following
diffs.
Reviewed By: michalgr
Differential Revision: D4021490
fbshipit-source-id: 517fd9033051c11ba97d312b16382445ae85d3f3
Summary:
**Motivation**
I'm working on a project that uses React Native and needs to add direct synchronous bindings to native stuff through the JavaScriptCore C API. This is because it's performance-sensitive and would benefit from the quickest JS->C path. It does this using cross-platform C++ code that works on both iOS and Android. Most of the infrastructure for getting access to the JSC context is already in React Native actually, just had to add a few more things.
(lexs you mentioned to tag you in this pull request)
**Test plan**
Modify the JavaScriptCore context through the `JSContextRef` returned (eg. add an object at global scope) and verify that it exists in JavaScript.
Closes https://github.com/facebook/react-native/pull/10399
Differential Revision: D4080945
Pulled By: lexs
fbshipit-source-id: 6659b7a01e09fd84475adde183c1d3aca2d4cf09
Summary: Instead of sending a list of modules over to JS on startup (and actually blocking script execution) instead provide a proxy object that constructs each of these lazily.
Reviewed By: lexs
Differential Revision: D3936979
fbshipit-source-id: 71bde822f01eb17a29f56c5e60e95e98e207d74d
Summary: Get rid of the old behaviour of JSON encoding in `nativeRequireModuleConfig` and consistently use the same names for function types "async/promise/sync"
Reviewed By: lexs
Differential Revision: D3819348
fbshipit-source-id: fc798a5abcaf6a3ef9d95bd8654afa7825c83967
Summary:
JSC's utf16 -> utf8 conversion crashes on encountering bad utf16. Instead, use our own conversion (conveniently copied from fbjni).
Original fix thanks to rigdern (https://github.com/facebook/react-native/pull/9302)
Reviewed By: mhorowitz
Differential Revision: D3746947
fbshipit-source-id: 29887ca720f6a2b074f01f853bad28a083b273bc
Summary:
We have a lot of small-ish calls to JSC and within the bridge that add up during TTI. This gives us a way to measure them in aggregate in a reasonable way.
From the comments:
MicroProfiler is a performance profiler for measuring the cumulative impact of
a large number of small-ish calls. This is normally a problem for standard profilers
like Systrace because the overhead of the profiler itself skews the timings you
are able to collect. This is especially a problem when doing nested calls to
profiled functions, as the parent calls will contain the overhead of their profiling
plus the overhead of all their childrens' profiling.
MicroProfiler attempts to be low overhead by 1) aggregating timings in memory and
2) trying to remove estimated profiling overhead from the returned timings.
To remove estimated overhead, at the beginning of each trace we calculate the
average cost of profiling a no-op code section, as well as invoking the average
cost of invoking the system clock. The former is subtracted out for each child
profiler section that is invoked within a parent profiler section. The latter is
subtracted from each section, child or not.
The usage is similar to Systrace: you put a MICRO_PROFILER_BLOCK in the block you want to profile and C++ RAII will handle timing it.
After MicroProfiler::stopProfiling() is called, a table of tracing data is emitted
to glog (which shows up in logcat on Android).
Differential Revision: D3635319
fbshipit-source-id: 01390b8ac76a68dd425cba2adfdde6e4957440cc
Summary: Add a new interface to JSC that allows loading a file lazily from disk, i.e. using mmap, instead of loading the whole file upfront and copying into the VM.
Reviewed By: michalgr
Differential Revision: D3534042
fbshipit-source-id: 98b193cc7b7e33248073e2556ea94ce3391507c7
Summary: JSCExecutor has something like this, but for methods on JSCExecutor itself. Open to better ideas around how to share code
Reviewed By: lexs
Differential Revision: D3516314
fbshipit-source-id: 4b1265916c52d582bb0b9348e9b4a099f566d6c9
Summary:
Adds:
- getType so you can switch on type
- getPrivate
Reviewed By: lexs
Differential Revision: D3515510
fbshipit-source-id: d574b04f563ac650bacec3751b50be6345e8439a
Summary:
Adds:
- callAsConstructor
- Ability to provide `this` object to callAsFunction
- getPropertyNames() now returns Strings (which can be converted by the caller to std::string if they want).
Fixes:
- double free issue with the String move constructor
Reviewed By: lexs
Differential Revision: D3515398
fbshipit-source-id: afa1342044e41fdd833dd27b8a244a58d4078442
Ashok Menon
Michael Bolin
Charles Dick
Pieter De Baets
Mathieu Baudet
Alexey Lang
Sean Kinsey
Marc Horowitz
Dan Caspi
Stephen Girdlestone
Michał Gregorczyk
Michael Lee
David Aurelio
Lukas Piatkowski
Alexander Blom
Ben Nham
Giuseppe Ottaviano
Nikhilesh Sigatapu
Scott Wolchok
Emil Sjolander
leeight
Hoa Dinh
Chris Hopman
Adam Ernst
Kellie Medlin
Andy Street
Tadeu Zagallo
Clement Genzmer
Greg Pstrucha
Mark Wang