| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
They were using "<ATTR>" which was parsed as a HTML tag.
ctx.file appeared as:
It is a shortcut for:
list(ctx.attr..files)[0]
ctx.files appeared as:
It is a shortcut for:
[f for t in ctx.attr. for f in t.files]
--
MOS_MIGRATED_REVID=119149347
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=119141878
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This halves the overhead with sandboxing enabled vs disabled for a test
that basically only mounts a bunch of files out of a directory, and
slows that same test with a single extra file added to the directory
(but not mounted) by only ~4%.
The test is <https://gist.github.com/bsilver8192/10527a862ce16bb7f79a>
with 30000 inputs moved to a subdirectory and only 10 genrules.
This change means symlinks will be mounted directly as their target
rather than as a symlink, but this solves some weird behavior with
multi-level symlinks and will only break things which don't declare all
of their dependencies.
--
Change-Id: I1aa39dccb2e5fca2893bdab9065ee043d34019b2
Reviewed-on: https://bazel-review.googlesource.com/#/c/3220/
MOS_MIGRATED_REVID=119138157
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously AspectFunction was using an Aspect from the SkyKey, which
might have been stale.
This CL fixes the bug as uncovered in the test (see SkylarkAspectsTest),
but further refactoring is needed since SkylarkAspectClass equals() is
incorrect, and in fact obtaining the Skylark aspect definition should
always introduce Skyframe dependency.
--
MOS_MIGRATED_REVID=119137636
|
|
|
|
|
|
|
|
|
|
|
| |
There is no GetInstallDir() function that populates |extracted_binaries|
vector. The one that exists and populates it, is called GetInstallBase()
in src/main/cpp/blaze.cc.
--
Change-Id: I09f951a132800e133197f5ec3206f436f90555f9
Reviewed-on: https://bazel-review.googlesource.com/c/3250/
MOS_MIGRATED_REVID=119107145
|
|
|
|
|
|
| |
--
Reviewed-on: https://github.com/bazelbuild/bazel/pull/1119
MOS_MIGRATED_REVID=119104941
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=119104560
|
|
|
|
|
|
|
| |
when copying.
--
MOS_MIGRATED_REVID=119094214
|
|
|
|
|
|
|
| |
question, rather than during loading of the target's package. This way a target's package won't be in error if e.g. an unrelated target has empty 'srcs'.
--
MOS_MIGRATED_REVID=119079777
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=119055012
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=119053259
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=119051781
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 5d737d642623: ("Some minor updates to the BUILD files.") made the
default visibility for the whole package public, and thus setting it
again in blaze_util target is rendundat.
While at it remove blaze_util_platform.h from srcs attribute of client
target, as it is already in hdrs list of blaze_util target and client
already depends on it, and thus it is not necessary to list it again
here.
--
Change-Id: I995c3cd6e2f63384acef4b24b9a082e7cb59331a
Reviewed-on: https://bazel-review.googlesource.com/#/c/3150/1
MOS_MIGRATED_REVID=119041634
|
|
|
|
|
|
|
| |
--
Change-Id: Ib83af0d0a04dc6b173bef1df28d17abc7a3c824d
Reviewed-on: https://bazel-review.googlesource.com/#/c/3120/
MOS_MIGRATED_REVID=119027507
|
|
|
|
|
|
|
|
| |
Resources in `java_library`s get the external repo path added to them, e.g., `foo` becomes `external/git_remote_repo/foo`. It needs to be removed, though I'm not sure this is the right way to strip it ...
--
Reviewed-on: https://github.com/bazelbuild/bazel/pull/1085
MOS_MIGRATED_REVID=119024770
|
|
|
|
|
|
|
| |
Then less special-casing is needed and these files are always read by the compiler anyway, so we don't lose any incrementality.
--
MOS_MIGRATED_REVID=119023956
|
|
|
|
|
|
|
|
|
| |
building watch OS 1 apps.
RELNOTES: Support apple_watch1_extension and apple_watch_extension_binary rules for creating watch OS 1 extensions.
--
MOS_MIGRATED_REVID=119000703
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118969066
|
|
|
|
|
|
|
|
|
|
|
| |
android_binary (and java_binary).
RELNOTES[NEW]: The output of -printconfiguration of proguard is available
through an implicit "[target_name]_proguard.confg" output of android_binary (and
java_binary).
--
MOS_MIGRATED_REVID=118958246
|
|
|
|
|
|
|
| |
native rules to skylark rules and that returned from skylark rules to depending native rules. This fixes a bug where a skylark rule that did not touch the objc provider would automatically pass up the objc provider it received from dependencies to dependant native rules, leading to duplicate symbol linking errors when the same objc provider was exported twice.
--
MOS_MIGRATED_REVID=118947782
|
|
|
|
|
|
|
| |
names contain url, http or https.
--
MOS_MIGRATED_REVID=118946808
|
|
|
|
|
|
|
|
|
|
|
| |
In the experimental UI, use progress events as an occasion to
update the progress bar. However, to avoid unnecessary noise, only
do so, if the progress bar potentially changes with time.
--
Change-Id: I9bee093c29e0eb50e7fcce2fa6e941a02d615def
Reviewed-on: https://bazel-review.googlesource.com/#/c/3206
MOS_MIGRATED_REVID=118944208
|
|
|
|
|
|
|
|
|
| |
1) Introducing the action_config message in the crosstool protobuf definition. The only part of that definition that are implemented in this CL is the "tool" section, other parts will be implemented in future CLs. The proto fields are here now to avoid being delayed by release cycles at each step of the implementation.
2) Refactoring the implementation of the "feature algebra" that computes the enabled features for a given toolchain. An interface called "CrosstoolActivatable" is used to represent any participant int the feature algebra, and can be either a feature or an action_config.
--
MOS_MIGRATED_REVID=118943663
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118943054
|
|
|
|
|
|
|
|
|
| |
Fixed #1040
RELNOTES:none
--
MOS_MIGRATED_REVID=118941667
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118940625
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently, the progress bar shows the 3 earliest started still running actions.
While this might give some indication about which actions delay the build process,
it lacks quantitative information about how long the delay is. By adding the current
run time to the progress bar, at least some quantitative information is provided.
--
Change-Id: Ib90eda67d87384a4372cda7779a09a44cea2b8dd
Reviewed-on: https://bazel-review.googlesource.com/#/c/3205
MOS_MIGRATED_REVID=118933973
|
|
|
|
|
|
|
|
|
|
| |
In this way, the ExperimentalStateTracker has access to all the information needed
to provide information about the run-time of currently running actions.
--
Change-Id: I0f4e48f39e9ebe63555e4bb1d70df2a6dbb65430
Reviewed-on: https://bazel-review.googlesource.com/#/c/3204
MOS_MIGRATED_REVID=118929758
|
|
|
|
|
|
|
|
|
|
|
| |
special-casing it in CppConfiguration.
This seems to be the most reasonable solution. I was toying with the idea of adding a field to CROSSTOOL but that would fail if you set libc_top to something other than what was specified in that file. If I had a infinite amount of time, I'd create a custom rule called cc_libc where libc_top would point so that this file can be referenced by an attribute, but since I don't, this seems to be workable compromise.
Also note that contrary to what you'd glean from the code, we don't actually have "compile" and "link" filegroups for libc.
--
MOS_MIGRATED_REVID=118921101
|
|
|
|
|
|
|
| |
Fixes #1087
--
MOS_MIGRATED_REVID=118919990
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118819625
|
|
|
|
|
|
|
|
|
| |
The compiler may implicitly load types that are direct dependencies but not
referred to in source, e.g. in the case of package-info files where the package
symbol is implicitly loaded to resolve annotations.
--
MOS_MIGRATED_REVID=118800795
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118786316
|
|
|
|
|
|
|
| |
BuildConfiguration.Fragment is much more complicated than it should be. Every little helps.
--
MOS_MIGRATED_REVID=118783105
|
|
|
|
|
|
|
|
|
| |
memory usage
Label instances do not share a single intern pool today, as some other objects, such as SkyKey and PackageIdentifier. This change allows all Label instantiations to use a single intern pool for better memory usage.
--
MOS_MIGRATED_REVID=118780952
|
|
|
|
|
|
|
|
|
| |
Noticed this while messing with remote runfiles.
RELNOTES: local_repository can take a relative path.
--
MOS_MIGRATED_REVID=118780775
|
|
|
|
|
|
|
|
|
| |
I was messing with filesystem dirtiness checkers and got an NPE, and I guess it
was so early in the build that the buildRequest didn't have WorkerOptions yet,
so I got a NPE from WorkerModule.buildComplete masking the actual NPE.
--
MOS_MIGRATED_REVID=118780059
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118727286
|
|
|
|
|
|
|
| |
Fixes #855.
--
MOS_MIGRATED_REVID=118711400
|
|
|
|
|
|
|
| |
is a no-op by default.
--
MOS_MIGRATED_REVID=118705742
|
|
|
|
|
|
|
| |
code) if we actually crashed.
--
MOS_MIGRATED_REVID=118702463
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118699141
|
|
|
|
|
|
|
|
| |
This is problematic for coming up with a less stateful BlazeModule API, which
- in turn - is a requirement for running multiple commands in the same server.
--
MOS_MIGRATED_REVID=118696892
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
--java_langtools=@bazel_tools//tools/jdk:langtools
--javabuilder_top=@bazel_tools//tools/jdk:JavaBuilder_deploy.jar
--singlejar_top=@bazel_tools//tools/jdk:SingleJar_deploy.jar
--genclass_top=@bazel_tools//tools/jdk:GenClass_deploy.jar
--ijar_top=@bazel_tools//tools/jdk:ijar
--javac_bootclasspath=@bazel_tools//tools/jdk:bootclasspath
--javac_extdir=@bazel_tools//tools/jdk:extdir
--java_toolchain=@bazel_tools//tools/jdk:toolchain
--host_javabase=@bazel_tools//tools/jdk:jdk
--javabase=@bazel_tools//tools/jdk:jdk
--
MOS_MIGRATED_REVID=118692509
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118683261
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Treat SUBCOMMAND events the same way as ERROR/WARNING/INFO commands in the
experimental UI: they provide information the user wants to look up later.
As SUBCOMMAND events are only raised if that information is requested by the
user, providing this information does run the risk over overwhelming the user
with information.
--
Change-Id: Ia7adeb2d5b92f18b8b88b026f790ac2af47b59d4
Reviewed-on: https://bazel-review.googlesource.com/#/c/3201
MOS_MIGRATED_REVID=118678085
|
|
|
|
|
|
|
| |
Constants.ANDROID_DEFAULT_FAT_APK_CPUS.
--
MOS_MIGRATED_REVID=118675987
|
|
|
|
|
|
|
|
|
| |
This is conceptually better (because each fragment should be
self-contained), and gives us better performance, as we no longer need to
load all explicit labels.
--
MOS_MIGRATED_REVID=118674470
|
|
|
|
|
|
|
|
|
|
| |
compilation inputs. As
of r264664, Clang doesn't require files in module maps to be present anymore if
the sources are instead available in the corresponding .pcm files.
--
MOS_MIGRATED_REVID=118670841
|
|
|
|
|
|
|
| |
Constants.DEFAULT_PACKAGE_PATH.
--
MOS_MIGRATED_REVID=118666575
|