Today both `mtouch` and `mmp` are copying the entire `.framework`
directories inside the `.app\[Contents\]Frameworks\` directory.
However not everything in a framework is required at runtime. The most
common unrequired files would be headers (`Headers/*.h`) and modules
(`Modules/*`).
Looking at Xcode build output we can see something like:
```
builtin-copy -exclude .DS_Store -exclude CVS -exclude .svn -exclude .git -exclude .hg -exclude Headers -exclude PrivateHeaders -exclude Modules -exclude \*.tbd -bitcode-strip replace-with-marker -bitcode-strip-tool
```
which excludes a few more, less common, files.
This _builtin_ command is not available externally (for us to re-use)
but it hints that Xcode is likely using `rsync` to avoid copying part of
the files.
Note: the builtin command also _likely_ calls `bitcode_strip` too (or has
similar code embedded) and `mtouch` already does so too
There's a cost to spawning an external process, like `rsync`, which we
avoid by having our own file copier, which clones files (almost zero
cost). That does not support excluding files, but deleting files is also
very cheap. Testing shows copying a framework to be less than 1 ms, even
with with extra deletion step.
* Tweak `GetRealPath` to optionally not to warn if the path does not exists
since, in this case, it's a check we want to do after resolving the path
This fixes several (5) MTouch tests looking for specific (and no extra)
warnings
```
Unable to canonicalize the path '/Users/builder/azdo/_work/2/s/xamarin-macios/tests/mtouch/bin/Debug/tmp-test-dir/Xamarin.Tests.BundlerTool.CreateTemporaryDirectory195/testApp.app/Frameworks/Mono.framework/CVS': No such file or directory (2).
```
because:
> If the file to be deleted does not exist, no exception is thrown.
https://docs.microsoft.com/en-us/dotnet/api/system.io.file.delete?view=net-5.0
and yes this is different from `Directory.Delete` and PR https://github.com/xamarin/xamarin-macios/pull/10441
* Fix MT0015 test failure
The MT0015 test creates a directory where a file is expected.
That's fine except the code for handling this was a bit weird. It
worked because of a `TryDelete` on the path, which avoided the
`UnauthorizedAccessException` when `File.Delete` is used on a path.
Then later there's a `Directory.Exists` check that would throw...
The code now does the `Directory.Exists` first and only call
`File.Delete` if it returns false. The throwing of the exception
is kept since the code (and test) are already present (and this
minimize changes and chance of other surprises)
* [tests] Build test-libraries for Mac Catalyst.
* [msbuild] Add support for Mac Catalyst binding projects.
* [mtouch] Allow frameworks for Mac Catalyst apps.
* [mtouch] Put frameworks in the expected location for Mac Catalyst apps.
* [msbuild] Create the Resources directory before trying to put files in it.
In some places we have to provide the macOS version, and in other places the
iOS version. Add a map and the corresponding code to convert between the two,
and use them when needed.
* [watchOS] Add x86_64 simulator support
* Build runtime/registrar x86_64 slices
* Produce a 64 bit version of Xamarin.WatchOS.dll
* Allow building x86_64 for watch simulators in mtouch
* Let xharness know about x86_64
* [tests] Add x86_64 arch to test-libraries
* Make dotnet package aware of x64
* [ObjCRuntime] Fix computing if we're calling a stret function or not in a 64-bit watchOS simulator.
* [xharness] Re-enable some watchOS tests.
Co-authored-by: Rolf Bjarne Kvinge <rolf@xamarin.com>
Some appextension mtouch code had to be moved to shared code. This code is currently
only used for iOS/tvOS/watchOS, but it will eventually be applicable to macOS as
well.
This makes it possible to re-use the registrar code in dotnet-linker.
Fixes these linkall tests:
Linker.Shared.OptimizeGeneratedCodeTest
[FAIL] IsARM64CallingConvention : optimized: no ldsfld instruction
Expected: 0
But was: 1
at Linker.Shared.BaseOptimizeGeneratedCodeTest.IsARM64CallingConvention() in /Users/rolf/work/maccore/main/xamarin-macios/tests/linker/BaseOptimizeGeneratedCodeTest.cs:line 527
[FAIL] SetupBlockPerfTest : At least 6x speedup
Expected: greater than 6
But was: 1.0876440665344851d
at Linker.Shared.BaseOptimizeGeneratedCodeTest.SetupBlockPerfTest() in /Users/rolf/work/maccore/main/xamarin-macios/tests/linker/BaseOptimizeGeneratedCodeTest.cs:line 120
And linkall is now green for .NET/Debug.
Refactor the Optimizations class to have no conditionally compiled code, which makes
it re-usable from our dotnet-linker code.
Also return any errors or warnings instead of showing/throwing them, which makes
the caller able to show them using whatever means is easiest for the caller.
One test needed an update to the list of valid optimizations, because we now have
a per-platform map of valid optimizations, instead of just a iOS/tvOS/watchOS vs
macOS split ('remove-unsupported-il-for-bitcode' is only valid for watchOS, and now
we say so, while we previously said it was a valid optimization for iOS and tvOS
as well, even though we'd warn about it and do nothing if you tried to set it).
* [mmp] Rename LinkMode.All to LinkMode.Full.
So that we can continue to use Enum.Parse<LinkMode> to parse 'Full' as the link mode.
* [dotnet] Implement support for our different link modes.
Tell the managed linker what to do with each input assembly depending the selected
link mode (link all, link sdk, don't link).
Refactor Application creation to happen earlier, and to split out the cache
creation. This way we can create the Application instance before processing
the configuration, and as we process any configuration we can set properties
on the Application instance.
If we can't find the mscorlib assembly in the list of loaded assemblies, try to load
it explicitly. If we still can't find it the mscorlib assembly, look for System.Void
in any assembly. This shouldn't be a performance bottleneck, because we cache the
System.Void type, which means the lookup is only done once.
This makes System.Void lookup work when building with .NET as well, since there's
no mscorlib.dll there.
This is required when running mtouch and mmp to generate the partial static registrar
code for .NET.
Also add a 'None' build target for the BuildTarget enum for when we're
building for neither simulator nor device (i.e. macOS). This means the default
value will change (since 'Simulator' is no longer the first value), but as far
as I can tell we're always assigning a specific value and not relying on the
default, so this should not make any difference.
This will be needed when the .NET code starts using these classes.
A few changes are required to have an Application instance at hand when we need to
get the ProductName from it.
This is necessary for .NET, since there will be a single linker library for all platforms,
which means we can't use a constant.
* [mtouch/mmp] Share Application.IsDualBuild, Is32Build and Is64Build.
* [mtouch/mmp] Share --tls-provider and --http-message-handler.
* [mtouch/mmp] Share --force.
* [mtouch/mmp] Share --cache.
* [mtouch/mmp] Share --nolink, --linksdkonly, --linkplatform and --linkskip.
* [mtouch/mmp] Share --i18n.
* [mtouch/mmp] Share --xml.
* [mtouch/mmp] Share --registrar and --runregistrar.
* [mtouch/mmp] Share --warn-on-type-ref.
* [mtouch/mmp] Share --sdk.
* [mtouch/mmp] Share --debug.
* [mtouch/mmp] Share --reference, and deprecate -r|--ref and -a|--assembly.
* [mtouch/mmp] Share --targetver, and deprecate mmp's --minos.
* [msbuild] Adjust tests after switching to use --reference instead of -r.
* Update according to review.
* [mmp] Remove --registrar:il.
The IL generator was what MonoMac had before the dynamic/static registrar code
got shared between MonoTouch and MonoMac. The IL registrar been gone for
years, and as far as I know nobody ever used --registrar:il, even though it
was provided as a compatibility option in the beginning (we still had the IL
registrar around for a while after adding the static+dynamic registrars, until
it was completely replaced by the dynamic registrar).
So just remove this option, if anyone ever used it they can replace it with
--registrar:dynamic.
* [mtouch/mmp] Keep bundler-specific code in its corresponding file.
Rolf Bjarne Kvinge
Sebastien Pouliot
Alex Soto