Creating a new NSString doesn't always lead to creating a new NSString, which
will obviously cause trouble.
The scenario is:
* An NSString with the value @"Bye" is added to an NSDictionary, with the same
string as both the key and the value.
* The (managed) string indexer is used to try to get the value back. The
string indexer would call 'new NSString ("Bye")', which would create a new
managed NSString, and maybe a new native NSString (or maybe it would re-use
an existing NSString). Then the handle of this NSString would be passed to
the native API, and the same handle would come back as the result (since the
same string is both the key and the value). We'd call Runtime.GetNSString on
the returned handle, get back the same managed instance that was created
just before the call to the native method. Finally, just before returning
this managed instance from the indexer, we'd dispose it... since it was
created in a 'using' block. Then we'd return a disposed NSString object from
the indexer. Ops.
The fix is to not create a managed wrapper for the NSString handle we need to
pass to the native API, but create and free the native NSString object without
using a managed wrapper.
I obsoleted `GetProjectPoint` in favor of `Project` because of the introduction of `Unproject` (which made me realize the naming was wrong) and based on the API doc https://developer.apple.com/documentation/arkit/arcamera/2923538-projectpoint?language=objc.
The `CGPoint` returned is the projection of a point. An other naming option would have been `GetProjectedPoint` but I think `Project` is closer to the original and clear enough. You project/unproject something onto something else and you get the projection back.
* Bump to use Xcode 10 beta 1
* Update Versions.plist
* Add a dependency on Xcode 9.4.
* [msbuild] Fix build with Xcode 10 beta 1. (#4182)
Many years ago (in Xcode 7 according to code comment)
Developer/Platforms/iPhoneOS.platform/Developer/usr disappeared, and we coped
by looking at Developer/usr instead (and also the subsequent code to locate
the bin directory was based on the location of the usr directory).
Developer/Platforms/iPhoneOS.platform/Developer/usr reappeared in Xcode 10
beta 1, but it seems useless (for one it doesn't contain a bin directory), so
in order to try to keep things sane don't look for this directory in Xcode 10
and instead go directly for Developer/usr (which is what we've been using as
the usr directory for years anyway).
Fixes this problem when building apps with Xcode 10 beta 1:
/Library/Frameworks/Mono.framework/External/xbuild/Xamarin/iOS/Xamarin.iOS.Common.targets(626,3): error : Could not locate SDK bin directory [/Users/rolf/Projects/TestApp/test-app.csproj]
* [runtime] Build 32-bit mac executables using Xcode 9.4.
* [mtouch] Work around broken tvOS headers in Xcode 10 beta 1.
* [mtouch] Work around build problem with Apple's simd headers in Objective-C++ mode.
* Use version-agnostic paths to sdk directories.
* [tests][xtro] Add todo files (from unclassified) and adjust ignore files to avoid errors
* [macos][security] Re-enable SSL[Get|Set]AlpnProtocols. Fixes#4001 (#4022)
* [macos][security] Re-enable SSL[Get}Set]AlpnProtocols. Fixes#4001
This was fixed in macOS 10.13.4
https://github.com/xamarin/xamarin-macios/issues/4001
* [tests][monotouch-tests] Disable a few test cases (one crasher, other failures). Causes to be verified later
* [xharness] Fix permission dialog suppression in Xcode 10.
* [xharness] Ignore 32-bit macOS tests by default.
* [tests] Execute mmp regression tests with Xcode 9.4 since many of them are 32-bit and needs porting to 64-bit.
* [mmptest] Ignore 32-bit XM tests if we don't have a 32-bit-capable Xcode.
* [registrar] Add workaround for broken headers in Xcode 10 beta 1 (radar 40824697).
* [mtouch] Restrict another workaround for an Xcode 10 beta 1 bug to a specific Xcode version to remove it asap.
* [tests] Fix some protocol changes (public or not) find by introspection tests
* [tests][intro] Fix DefaultCtorAllowed failures
* [Intents] Obsolete several Intents classes in watchOS.
Several existing Intents classes have been marked as unavailable in watchOS in
the headers in Xcode 10 beta 1, and corresponding tests are now failing.
So obsolete the managed wrapper types, and fix tests accordingly.
* Fix xtro wrt previous Ietents/intro changes
* [tests] Minor adjustments to mtouch tests to work with Xcode 10.
* [msbuild] Update tests to cope with additional files produced by the Core ML compiler.
* [msbuild] Xcode 10 doesn't support building watchOS 1 apps, so show a clear error message explaining it.
Also update tests accordingly.
* [coreimage] Stub new filters and exclude ?removed? ones from tests
* Update GameplayKit and SpriteKit NSSecureCoding _upgrade_ and fix other non-public cases (in tests)
* [tests] Ignore some GameKit selectors that don't respond anymore (but seems to be available, at least in header files)
* [tests] Fix intro 32bits testing for filters resutls
* [msbuild] Slightly change error message to be better English.
* [monotouch-tests] Adds mono's WeakAttribute tests
* Embrace watchOS
Unfortunately this revealed that WeakAttribute is not working for watchOS
* Port mono's WeakAttribute test to our test runner
Test Ported:
5bdaef7e5f/mono/tests/weak-fields.cs
* Fix object leaks and implement suggested approach from 4b9ade0c59
* Remove debug spew and fix formating on header
First there was darkness; no blocks were retained.
Then came the light; and all blocks were retained [1]
Forever.
But all that once is, must one day not be,
and thus the light gave way to darkness,
and blocks were only retained as long as need be [2].
But before there was a balance, there was a crossroad.
In some places the light shone forever,
and all blocks were retained.
In other places there was a balance,
and the light shone only as long as needed.
A desire to unify arose.
Alas, it could not be.
It was a bright and sunny day
When a merge failed [3].
And all blocks were retained. Twice.
Once [here][4] and once [there][5].
For many years we could not see.
Until a dark and rainy night,
when an awareness arose.
And the desire to unify the balance could finally be fulfilled.
[1]: 6efca92acb
[2]: a22f877539
[3]: befa0477cf
[4]: 5158a3c001/src/ObjCRuntime/Runtime.cs (L858)
[5]: 5158a3c001/runtime/runtime.m (L2091)
* [monotouch-test] Stop MidiThruConnectionTests.FindTest from randomly failing (#3696)
Fixesxamarin/maccore#658
When a MidiThruConnection is created but for some reason is not disposed
the system keeps it alive even between app/simulator restarts, if you want
to clean the connections you must reset contents and settings from simulator.
In order to avoid this test from failing randomly and since the intent of the
test is to check if `MidiThruConnection.Find` works we change the assert to
`>= 2` since this is at least the number of connections we expect.
* [monotouch-test] Enhance MidiThruConnectionTests.FindTest from #3696 comments (#3700)
Fixesxamarin/maccore#658
When a MidiThruConnection is created but for some reason is not disposed
the system keeps it alive even between app/simulator restarts, if you want
to clean the connections you must reset contents and settings from simulator.
Which is a bit harder when the issues happens on the macOS.
* [security] Modifying structure of bindings
Added strong dictionary for key generation according to
https://developer.apple.com/documentation/security/certificate_key_and_trust_services/keys/key_generation_attributes
in preparation making a strongly typed key generation possible.
* Making strong dictionary and composite of other strong dictionaries.
* Implementing access control as property type of SecPublicPrivateKeyAttrs.
* Adding new overload for SecKey.CreateRandomKey.
* Moving TokenID to strongly typed property.
* Fix coding style + use nameof
* Fixing Xcode version assertion of key generation tests.
* Fixing errors in test case.
* Fixing whitespace issue.
* Resolving inheritance issue
Removing inheritance between SecKeyGenerationParameters and
SecPublicPrivateKeyAttrs and instead add the necessary properties
to SecKeyGenerationParameters. Adds redundant code, but I don't
see a way around it.
* Moving test case to appropriate test class.
* Creating necessary strong dictionaries for key generation.
* [formatting] Mono coding guidelines
Make sure you're following http://www.mono-project.com/community/contributing/coding-guidelines/
In VSMac: Preferences > Source Code > Code Formatting > C# source code > Policy Mono.
* Remove [Advice] that are specific to GenerateKeyPair
`SecKeyGenerationParameters` and `SecKeyParameters` cannot be used for `GenerateKeyPair`.
* Clarify 'ArgumentException' for invalid 'SecKeyType'
* Fixed CreateRandomKeyTest
- Renamed CreateRandomKeyWithParametersTests to CreateRandomKeyTest.
- Add messages to all asserts.
```
keyGenerationParameters = new SecKeyGenerationParameters ();
keyGenerationParameters.KeyType = SecKeyType.Invalid;
Assert.Throws<ArgumentException> (() => { SecKey.CreateRandomKey (keyGenerationParameters, out _); }, "CreateRandomKey - invalid key type");
```
- ^ didn't work because `SecKeyGenerationParameters`'s setter protects against invalid. There's no constant for invalid so null is returned.
```
Assert.That (SecKey.CreateRandomKey (keyGenerationParameters, out _), Is.EqualTo (SecStatusCode.Param), "CreateRandomKey - Param issue, invalid RSA key size");
```
- ^ `SecKey.CreateRandomKey` doesn't return a `SecStatusCode` like `GenerateKeyPair`.
* Fixes based on spouliot's input.
* Mono styling fix.
Implement support for optimizing away the dynamic registrar.
* Add a new property (Runtime.DynamicRegistrationSupported) that indicates at
runtime whether dynamic registration is available.
* Check this new property whenever we need dynamic registration.
* Add an optimization to mtouch/mmp that detects whether dynamic registration
is required, and if not, changes Runtime.DynamicRegistrationSupported to
return false.
* Add an optimization to mtouch/mmp that inlines
Runtime.DynamicRegistrationSupported as a constant value.
The end result is that the linker will link away the dynamic registrar if
mtouch/mmp detects that it's not needed.
Benchmark
---------
I've compared the size of entire apps built for device:
|test | Before | After | Diff | % |
|:-----------------------------|-------:|-------:|-------:|------:|
|[monotouch-test/Release][1] | 95.7mb | 95.0mb | -680kb | -0.7% |
|[link sdk/Release][2] | 21.2mb | 20.9mb | -245kb | -1.2% |
|[minimalistic app/Release][3] | 4.58mb | 4.32mb | -259kb | -5.7% |
[1]: https://gist.github.com/rolfbjarne/3871e36d2de8db1a8eee1d9f9276d3d2#monotouch-test-release-wall-optimizations-enabled
[2]: https://gist.github.com/rolfbjarne/3871e36d2de8db1a8eee1d9f9276d3d2#link-sdk
[3]: https://gist.github.com/rolfbjarne/3871e36d2de8db1a8eee1d9f9276d3d2#minimalistic-app
Add a 'register-protocols' optimization that:
Improves static registrar to:
* Generate a new table of protocol -> managed wrapper type. This is required
to find the wrapper type without having the `[Protocol]` attribute around.
* Make the generated code implement protocols from [Adopts] attributes. This
makes it possible to link away the `[Protocol]` attribute, because the
native implementation of `conformsToProtocol:` does the right thing (we
might even be able to link away our complete `ConformsToProtocol` logic when
we remove the dynamic registrar).
Improves linker to:
* Not mark protocol interfaces by the mere virtue of having a type that
implements them. This is implemented by not marking protocol interfaces when
they're implementing a class, but instead when a method implementation is
found to implement a method from a protocol interface.
* Mark the wrapper type for protocols (this allows us to remove the Protocol
attribute, since that's the link between the protocol and its wrapper type).
* Remove the [Protocol], [ProtocolMember] and [Adopts] attributes (but only if
optimizing protocols).
The static registrar still needs some of the information linked away, so a few
changes are required to make it available post linker.
Benchmark
---------
I've compared the size of entire apps built for device:
|test | Before | After | Diff | % |
|:-----------------------------|-------:|-------:|-------:|------:|
|[monotouch-test/Debug][1] | 101mb | 100mb | -888kb | -0.9% |
|[monotouch-test/Release][2] | 99.2mb | 95.4mb | -830kb | -0.9% |
|[minimalistic app/Debug][3] | 10.8mb | 10.4mb | -443kb | -4.1% |
|[minimalistic app/Release][4] | 4.7mb | 4.55mb | -157kb | -3.3% |
[1]: https://gist.github.com/rolfbjarne/0181ab8abe436c34cf4ee68ecfb8cd18#monotouch-test-debug
[2]: https://gist.github.com/rolfbjarne/0181ab8abe436c34cf4ee68ecfb8cd18#monotouch-test-release
[3]: https://gist.github.com/rolfbjarne/0181ab8abe436c34cf4ee68ecfb8cd18#minimal-xi-app-debug
[4]: https://gist.github.com/rolfbjarne/0181ab8abe436c34cf4ee68ecfb8cd18#minimal-xi-app-release
Create a table of interface -> protocol in the static registrar, since we need
to be able to look up a protocol given a managed type without looking at the
(possibly linked away) [Protocol] attribute.
* [tests] Build the native test library for macOS and create a binding project for it.
Also add the new binding project to the xammac and link all XM test projects,
which allows us to stop excluding tests that require the native library and
the corresponding bindings.
* [tests] Include more tests in xammac_tests.
* [tests] Correctly ignore the ObjC exception tests in release mode.
- Obsolete `ARFaceAnchor`'s default constructor because it's marked as unavailable.
- Obsolete `CreateFaceGeometry` in `ARSCNFaceGeometry` in favor of `Create` (same as `ARSCNPlaneGeometry`).
The original, now obsoleted, `LocalizedString` API returned a .net
`string` which does not work in most cases.
Different versions of iOS seems to return different (public or internal)
subclasses of `NSString` that are understood by other API (like NSString
`localizedStringWithFormat:`) for further customization.
Our logic to convert NSString to string is correct but it cannot
recreate the custom, required subclass to continue the localization.
So the new API return an `NSString` publicly (which is actually a
subclass) that can do the required job.
Adding a test in monotouch-test is presently blocked by #3265 [2]
[1] https://bugzilla.xamarin.com/show_bug.cgi?id=41292
[2] https://github.com/xamarin/xamarin-macios/issues/3265
* Add tests for new (and old) NSBundle API and adjust old ones since adding a Base.lproj directories changes things
* Add localization to xammac_tests since it shares the same, updated tests
and ensure the temporary `NSString` instance is disposed asap.
Found will debugging something else. Unit tests added to confirm there
is no behavior change when the API are used.
Fixesxamarin/xamarin-macios#3166
Apple broke `addAnimation:forKey:` API by changing `CAAnimation` parameter into a
`ISCNAnimationProtocol` that was introduced in Xcode 9 (iOS 11 and company).
We can't break the existing API but we can add an extension method that
reuses the `CAAnimation` overload but takes a `SCNAnimation` and we
turn the `SCNAnimation` into a `CAAnimation` behind the scenes.
For XAMCORE_4_0 we corrected the protocol signature to take `ISCNAnimationProtocol` so there won't be a need for the extension method anymore.
Largely related to somewhat recent frameworks/API additions to XM
ExternalAccessory was added to macOS 10.13 (Xcode9) and can't be run before that (including on the jenkins sierra bots). Also enable that test on tvOS (which added support in Xcode8)