| Commit message (Collapse) | Author | Age |
... | |
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124968182
|
|
|
|
|
|
|
| |
use_objc_header_names attribute.
--
MOS_MIGRATED_REVID=124955824
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124951254
|
|
|
|
|
|
|
|
|
| |
-stdlib=libc++.
RELNOTES: When building ObjectiveC++, pass the flag -std=c++11 in addition to -stdlib=libc++.
--
MOS_MIGRATED_REVID=124867581
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124848977
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124843113
|
|
|
|
|
|
|
| |
crosstool now points at appropriate clang in its action_configs.
--
MOS_MIGRATED_REVID=124830389
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Optional and named parameters are now specified using `defaultValue`
and `named` (and `positional`). The new structure allow for parameters
that are both named and positional (which was forbidden before).
This new structure will be used to give the @SkylarkCallable annotation
the possibility to provide named and default arguments. It should be
a functional no-op for Bazel users.
--
MOS_MIGRATED_REVID=124821455
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124815043
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124792198
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Breaks linking for all objc targets.
*** Original change description ***
Wrap inputs to ObjcLink (via clang) invocations in an objlist, and pass the args as a -filelist arg.
This prevents certain many-arg actions from making clang unhappy.
--
MOS_MIGRATED_REVID=124781451
|
|
|
|
| |
MOS_MIGRATED_REVID=124763523
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When true (the default), test rules included by --instrumentation_filter are instrumented. When false, test rules are excluded from instrumentation (whether or not their names would be matched by --instrumentation_filter).
This option provides a superior way to exclude test rules from instrumentation, compared to setting --instrumentation_filter exclusions based on test rule naming conventions. Naming conventions vary from language to language and are not always followed. Blaze already has enforced semantics for test targets (test rule build function names _must_ end in "_test", see TargetUtils.isTestRule). In general, when either option is available, it's better to rely on enforced semantics than unenforced conventions.
Note that this default preserves present behavior. In the future, I think it would make sense to:
* Change the default value of --instrument_test_targets to false.
* Change the default value of --instrumentation_filter to an empty string (match everything).
RELNOTES: Add --instrument_test_targets option.
--
MOS_MIGRATED_REVID=124732226
|
|
|
|
|
|
|
| |
--android_crosstool_top due to a busted checkArgument condition
--
MOS_MIGRATED_REVID=124616411
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, android_compiler would only be used if fat_apk_cpu and android_crosstool_top were both set. This led to confusing errors about invalid toolchains.
If --android_crosstool_top is set but --fat_apk_cpu is not, then --cpu will become --android_cpu during the transition. (android_cpu defaults to armeabi if it is not set.)
RELNOTES: If --android_crosstool_top is set, native code compiled for android will always use --android_compiler and not --compiler in choosing the crosstool toolchain, and will use --android_cpu if --fat_apk_cpu is not set.
--
MOS_MIGRATED_REVID=124611770
|
|
|
|
|
|
|
| |
tag's name attribute to the Instant Run application class.
--
MOS_MIGRATED_REVID=124606107
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124602618
|
|
|
|
|
|
|
| |
configurable or not.
--
MOS_MIGRATED_REVID=124565334
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
--
MOS_MIGRATED_REVID=124556360
|
|
|
|
|
|
|
|
|
| |
error when the LIPO context is not a cc_binary.
RELNOTES[NEW]: LIPO context (--lipo_context) can now also be a cc_test (in addition to cc_binary)
--
MOS_MIGRATED_REVID=124555465
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
--
MOS_MIGRATED_REVID=124553522
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
--
MOS_MIGRATED_REVID=124549894
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
--
MOS_MIGRATED_REVID=124548575
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124510100
|
|
|
|
|
|
|
| |
instead load it from //external.
--
MOS_MIGRATED_REVID=124480945
|
|
|
|
|
|
|
|
| |
The @SkylarkCallable annotation is very limited and to extend it
it is best to share a similar API than the @SkylarkSignature annotation.
--
MOS_MIGRATED_REVID=124473056
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124448791
|
|
|
|
|
|
|
|
| |
This fixes the breakage caused by commit 828a9e2f167db8c4debca6e00ee37dd030849c2f
Also modified test to check options for generating dependency file.
--
MOS_MIGRATED_REVID=124441471
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Makes it so module map generation action will be registered for all targets that should provide module maps. This allows other rules to depend directly on modulemap artifacts without turning on --experimental_objc_enable_module_maps for all builds.
* Disabled module maps for apple_binary because of a bug.
* objc_proto_library targets will register module map generation even with experimental union flag so that its modules can be imported by the dependent targets (even though the compilation is happening much later in the linking target).
RELNOTES:
--
MOS_MIGRATED_REVID=124386048
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124381114
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
AppleConfiguration.
Additionally, tweak single-architecture ios-platform logic such that ios_multi_cpus is checked before ios_cpu.
There are two contexts to note:
1. Single-architecture logic, (generally post-split), unaware of its own platform type aside
from configuration. This retrieves platform type from the --apple_platform_type configuration value.
a. getSingleArchPlatform() for Platform retrieval
b. getSingleArchitecture() for architecture retrieval
2. Multi-architecture logic, which should be aware of its own platform type, and passes it into
configuration accessors.
a. getMultiArchPlatform(PlatformType)
b. getMultiArchitectures(PlatformType)
All callers are migrated to these methods, though some still pass IOS platform type even though
they may need to be refactored to support additional platform types later.
--
MOS_MIGRATED_REVID=124370652
|
|
|
|
|
|
|
|
|
| |
args as a -filelist arg.
This prevents certain many-arg actions from making clang unhappy.
--
MOS_MIGRATED_REVID=124367172
|
|
|
|
|
|
|
| |
JavaLibraryHelper.
--
MOS_MIGRATED_REVID=124356892
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=124355936
|
|
|
|
|
|
|
|
|
| |
Also moved -S and -E options into the same action_config.
Since whoever consume CppCompileInfo add -c and -o options by themselves, to
keep compatible with this, the original code is still kept and used, until the corresponding flag sets are added into CROSSTOOL.
--
MOS_MIGRATED_REVID=124350905
|
|
|
|
|
|
|
|
|
|
| |
SkylarkSignature
Those arrays were checked to never have more than one value so use an empty name
instead of the empty array. This simplify comprehension of the API.
--
MOS_MIGRATED_REVID=124330918
|
|
|
|
|
|
|
|
| |
an action_config specifies no action, and is automatically applied to the
action being configured.
--
MOS_MIGRATED_REVID=124240929
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
experimental support
for building objc code using the c++ crosstool. This will eventually replace the current module support in blaze.
Note: This required injecting a CppModuleMap into the cc logic. The reason that objc cannot rely on standard CppModuleMap creation logic is that there is different naming semantics for module maps between cpp and objc. In particular:
- In cc, module maps can be inputs to a compilation action. Thus, the module maps are given labels.
- In objc, if interoping with swift, module maps are explicitly referenced in swift code. Thus, their names cannot contain illegal characters for swift source. Those, some name mangling occurs to get rid of "//" and ":". To enforce that this does not cause problems with compilation actions, the CPP_MODULE_COMPILE action has been disabled for the objc CcLibraryHelper.SourceCategory
--
MOS_MIGRATED_REVID=124177067
|
|
|
|
|
|
|
| |
isolate them from the linking target's defines and copts flags. Refactors CompilationAttributes into its own class with the Builder pattern.
--
MOS_MIGRATED_REVID=124137672
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Cannot be built with currently released Bazel.
*** Original change description ***
Remove redirect chasing from AndroidConfiguration.Loader.
With aliases, it's not necessary anymore.
--
MOS_MIGRATED_REVID=124123904
|
|
|
|
|
|
|
| |
With aliases, it's not necessary anymore.
--
MOS_MIGRATED_REVID=124112085
|
|
|
|
|
|
|
| |
Only getCompilationArtifacts() is used, which simply passes through a field, so this CL inlines it.
--
MOS_MIGRATED_REVID=124043284
|
|
|
|
|
|
|
|
|
|
|
| |
JavaCompilationArgsProvider instead of general deps.
In the current users of JavaLibraryHelper, emitProviders was always set to false using noProviders.
RELNOTES:
--
MOS_MIGRATED_REVID=124004929
|
|
|
|
|
|
|
| |
This fixes test timeouts when the app tested is using Swift. In theory this also prevents similar problems with any code that uses dynamic libs packaged in the IPA.
--
MOS_MIGRATED_REVID=123976581
|
|
|
|
|
|
|
|
|
| |
RELNOTES:
global function 'provider' is removed from .bzl files. Providers can
only be accessed through fields in a 'target' object.
--
MOS_MIGRATED_REVID=123956571
|
|
|
|
|
|
|
|
|
| |
The IDE currently extracts these from the output jars by
a fragile string comparison. This will allow the IDE to
do so more robustly (eg. in case the resource jar output is renamed).
--
MOS_MIGRATED_REVID=123939803
|
|
|
|
| |
MOS_MIGRATED_REVID=123909592
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=123868341
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=123858617
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=123851529
|