| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
|
| |
These are depended upon by analysis code, so need to live in the same library
as lib.analysis. Moving them here makes it possible to split the build-base
library into separate libraries for analysis, execution, and rules.
PiperOrigin-RevId: 164847161
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
RELNOTES: None
PiperOrigin-RevId: 164775289
|
|
|
|
|
|
|
| |
to get googletest.sh from to put in the runfiles.
RELNOTES: None
PiperOrigin-RevId: 164737919
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of having custom ArgvFragments for every combination of desired things, we make a combined "interpreter" of argvs. This saves memory and simplifies things as we do not have to allocate a strategy instance per call to args (instead pushing a single shared instance, followed by the args).
The generic interpreter does have a lot of branching compared to the bespoke implementations, but because the branch is always the same for long stretches the branch predictor should easily be able to handle it with minimal overhead (~1 cycle per branch IIRC).
This CL also elevates that we either want a NestedSet or an ImmutableCollection to the surface of the API, so consumers understand the cost when they call it with a non-immutable collection. Most of the changes in clients is due to this.
To cut down on CL churn, @Deprecated forwarding methods are added to CustomCommandLine. These will be removed in a separate CL using IDE inlining.
RELNOTES: None
PiperOrigin-RevId: 164725370
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Updated to avoid https://github.com/bazelbuild/bazel/issues/3501
This is a rollback of a rollback, with additional
modifications to BazelConfiguration.java to fix
https://github.com/bazelbuild/bazel/issues/3501,
the issue that was the reason we rolled back the
original change.
The additional updates serve to propagate the
client's TMP and TEMP envvars to the action, which
is a short-term solution to allow actions have a
TMP/TEMP envvar on Windows. They need at least one
of those to create temp directories.
The long-term solution is to set a value for TMP
or TEMP in the executor just before executing the
actions, so the TMP/TEMP would not be part of the
action key.
All of this only affects Bazel on Windows.
*** Original change description ***
Automated rollback of commit 0abf5fa2d64c76def5a8fa0f960b73ce0566af4d.
*** Reason for rollback ***
Breaks Bazel CI (https://github.com/bazelbuild/bazel/issues/3501)
*** Original change description ***
Android BusyBox: actions use the default shell env
SpawnActions that run the Android BusyBox now use
the default shell environment.
This has the following benefits:
- Bazel propagates the PATH, TMPDIR envvars to the
action
- Bazel propagates the --action_env envvars to the
action
This allows the Bazel client to pass
--action_env=TMP or --action_env=TEMP (whichever
of the envvars is defined) to the server, so the
BusyBox actions will have TMP/TEMP set (to the
same value as the clientenv), so they can create
temp directories using
java.nio.file.Files.createTempDirectory.
This method seems to be calling the GetTempPath
WinAPI function, which needs the TMP or TEMP
envvar, otherwise it falls back to returning
c:\windows which is non-writable.
There's one drawback of using the default shell
environment, although @ulfjack is working on it:
- PATH is now also part of the action's cache key.
However in a single-machine environment (no
remote execution) and assuming PATH isn't likely
to change between builds, this probably doesn't
poision the action cache in practice.
This change is a short-term solution. Propagating
the client env's TMP/TEMP means we make that part
of the action's cache key.
The ideal long-term solution will be to not
propagate this envvar, and instead let the
execution strategy set it to some
client-env-independent value.
See https://github.com/bazelbuild/bazel/issues/3264
Change-Id: I756a4203b5d86c881bc36cc089e35cde0d419914
RELNOTES: none
PiperOrigin-RevId: 164696600
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 164646858
|
|
|
|
|
|
|
|
|
| |
This is part of splitting up the build-base library into separate libraries for
analysis, exec, and rules. Unfortunately, there are a number of reverse deps
from analysis code to Skylark classes, so moving these is the only way to make
progress.
PiperOrigin-RevId: 164612957
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On Windows, Bazel will always use params files for
some BusyBox tools, because some flags of these
tools expect values with special characters in
them.
We need this change so Bazel can safely pass such
flags to the BusyBox on Windows.
See https://github.com/bazelbuild/bazel/issues/3264
RELNOTES: none
PiperOrigin-RevId: 164582899
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Change is needed due to the fact that java_import or other custom rules (genrules or Skylark) do not propagate coverage information. The lack of coverage data is caused by the fact that it is retrieved from compilation information and it is passed around through providers as Artifact(s) (also known as instrumentation metadata). The problem with the current implementation is that there is no way of retrieving instrumentation metadata from arbitrary jars provided by java_import or other custom rules.
The instrumentation metadata in the current implementation is a separate jar that contains uninstrumented classes for offline Jacoco instrumentation.
This change addresses that problem by having just one jar instead of 2 (the build jar and the instrumentation jar), adding the uninstrumented classes in the build jar and completely removing any other instrumentation metadata.
Implementation details:
* For each build jar there is a .txt file created that contains the relative path of each Java file. This file will also be included in the final build jar. It is used for recreating the correct path for each covered file when included in the coverage report.
* java_binary/java_test will set 2 environment variables:
1) JACOCO_METADATA_JARS - replaces the previous JACOCO_METADATA_JAR that was a jar that merged all the uninstrumented classes on the classpath in one jar. The new environment variable holds the paths of the runtime classpath jars - only some of them contain uninstrumented classes, letting the Jacoco coverage runner to filter and analyze them.
2) JACOCO_MAIN_CLASS - The main class to be called for the current coverage run. Previously this information was embedded in the JACOCO_METADATA_JAR's manifest.
PiperOrigin-RevId: 164562533
|
|
|
|
|
|
|
| |
This is part of splitting up the build-base library into separate libraries for
analysis, exec, and rules.
PiperOrigin-RevId: 164446955
|
|
|
|
|
|
|
| |
This is part of splitting up the build-base library into separate libraries
for analysis, exec, and rules.
PiperOrigin-RevId: 164438390
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Deprecate the --libraries flag of the
GENERATE_BINARY_R tool in favour of --library.
The new flag is multi-value and uses "," as the
pair-separator instead of ":".
The value converter still supports ":"-separated
pairs as well, but looks for "," first.
Old format:
--libraries=key1:value1,key2:value2,...
New format:
--library=key1,value1 --library=key2,value2
Motivation:
- the ":"-separator prevents using absolute paths
on Windows
The old flag is still supported, but will be
removed after 2018-02-28 (about 6 months from
now).
Also in this commit:
- add a new method to CustomCommandLine.Builder
to lazily construct the command line for the
--library flag
See https://github.com/bazelbuild/bazel/issues/3264
RELNOTES: none
PiperOrigin-RevId: 164246506
|
|
|
|
|
|
|
|
|
| |
Add SpawnAction.Builder#setProgressMessageNonLazy for dynamic strings. This should help guide users to the right method.
This also helped find a few methods I'd missed previously that could use a constant-string version.
RELNOTES: None
PiperOrigin-RevId: 164150264
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Breaks Bazel CI (https://github.com/bazelbuild/bazel/issues/3501)
*** Original change description ***
Android BusyBox: actions use the default shell env
SpawnActions that run the Android BusyBox now use
the default shell environment.
This has the following benefits:
- Bazel propagates the PATH, TMPDIR envvars to the
action
- Bazel propagates the --action_env envvars to the
action
This allows the Bazel client to pass
--action_env=TMP or --action_env=TEMP (whichever
of the envvars is defined) to the server, so the
BusyBox actions will have TMP/TEMP set...
***
PiperOrigin-RevId: 164126020
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
SpawnActions that run the Android BusyBox now use
the default shell environment.
This has the following benefits:
- Bazel propagates the PATH, TMPDIR envvars to the
action
- Bazel propagates the --action_env envvars to the
action
This allows the Bazel client to pass
--action_env=TMP or --action_env=TEMP (whichever
of the envvars is defined) to the server, so the
BusyBox actions will have TMP/TEMP set (to the
same value as the clientenv), so they can create
temp directories using
java.nio.file.Files.createTempDirectory.
This method seems to be calling the GetTempPath
WinAPI function, which needs the TMP or TEMP
envvar, otherwise it falls back to returning
c:\windows which is non-writable.
There's one drawback of using the default shell
environment, although @ulfjack is working on it:
- PATH is now also part of the action's cache key.
However in a single-machine environment (no
remote execution) and assuming PATH isn't likely
to change between builds, this probably doesn't
poision the action cache in practice.
This change is a short-term solution. Propagating
the client env's TMP/TEMP means we make that part
of the action's cache key.
The ideal long-term solution will be to not
propagate this envvar, and instead let the
execution strategy set it to some
client-env-independent value.
See https://github.com/bazelbuild/bazel/issues/3264
Change-Id: I756a4203b5d86c881bc36cc089e35cde0d419914
PiperOrigin-RevId: 164114502
|
|
|
|
|
|
|
|
|
|
| |
Consumers using spawn action builder now have access to handy overloads that behind the scene do a lazy String.format. In 95% of cases progress messages are expressible as 0, 1, or 2 argument String.formats.
This saves memory because the format string is constant and shared between all actions, and the captured subjects are usually live on the heap anyway (eg. labels).
Skylark still computes its progress messages eagerly. If we want similar savings there I'd have to follow up with a Skylark proposal.
PiperOrigin-RevId: 164068816
|
|
|
|
|
|
|
|
| |
Follows
https://docs.google.com/document/d/1aAIVWvHPERDz2cv_PCFGwr8dvh5FcAkENFoRsNS4clk/.
RELNOTES: None.
PiperOrigin-RevId: 163728291
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Actually hook up the resource filtering configuration transitions to
AndroidConfiguration's topLevelConfigurationHook.
Dynamic configuration is inefficient when multiple configurations are used.
Multiple configurations result in splitting the build graph and duplicating
work.
To avoid using multiple configurations as much as possible, dynamically
configured resource filtering will only be applied for top-level android_binary
targets. When android_binary targets are included as dependencies, it's very
likely that multiple binaries with many shared dependencies but different
configurations would be used. Only applying dynamic filtering to top-level
binaries removes this concern.
It is still possible to build multiple top-level binary targets with different
configurations at once. Previous versions of this code considered not using
dynamic configuration in that case as well, but that raised some unanswered
questions about whether Bazel's invariants should allow modifying one target's
configuration based solely on targets that happen to be building in parallel.
As a result, that optimization is not included for now; we assume that
developers manually building multiple similar binaries at once is relatively
rare (plus, dynamically configured resource filtering in general is not turned
on by default and will not be turned on by default until we're confident the
benefits outweigh the costs).
RELNOTES: none
PiperOrigin-RevId: 163464415
|
|
|
|
|
|
|
| |
Also fix the remaining violations.
RELNOTES: None.
PiperOrigin-RevId: 163391215
|
|
|
|
|
|
|
|
|
|
| |
This lets a parent choose a transition for its dep based
on the dep's rule class.
Implement (experimental) dynamic Android resource filtering
trimming with this.
PiperOrigin-RevId: 163259052
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Before this change, density-based resource filtering tracked resources by
qualifiers and name. Resources with density qualifiers specified would go into
this code, but only one resource would be chosen from each each (qualifier,
name) pair.
Instead, track the resource using its entire path, this tracking resources with
the same name seperately.
Also, in case multiple resource are passed to the resource processing action,
resource filtering only ignores a file if its name was in the list of resources
to ignore *and* it does not exist. Otherwise, legitimate resources with the
same name as a filtered resource might be ignored.
RELNOTES: none
PiperOrigin-RevId: 163235681
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 163126457
|
|
|
|
|
|
|
|
|
| |
Currently, allowable resource directories are defined in two places:
LocalResourceContainer, and android_ide_common (used, for example, by
ResourceFilter). Instead, keep them defined only in android_ide_common, a more authoritative source that we already rely on anyway.
RELNOTES: none
PiperOrigin-RevId: 163100497
|
|
|
|
|
|
|
| |
Also, a few small cleanups of some duplicate test methods.
RELNOTES: None
PiperOrigin-RevId: 163066349
|
|
|
|
|
|
|
| |
Fixes TODO in AndroidRuleClasses.java.
RELNOTES: None
PiperOrigin-RevId: 162955052
|
|
|
|
|
| |
RELNOTES: None
PiperOrigin-RevId: 162945309
|
|
|
|
| |
PiperOrigin-RevId: 162906919
|
|
|
|
|
|
|
|
| |
This flag has been a no-op for a while with a deprecation warning that it will
be removed.
RELNOTES: Remove --experimental_android_use_singlejar_for_multidex.
PiperOrigin-RevId: 162643922
|
|
|
|
|
|
|
| |
This is a no-op, the same classes inherited from them.
RELNOTES: None
PiperOrigin-RevId: 162627873
|
|
|
|
|
| |
RELNOTES: None
PiperOrigin-RevId: 162613289
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a partial rollback.
I restored the old flags (with PathListConverter),
added the new ones, added deprecation warnings and
annotations, and added code to concat the values
of new and old.
Existing released versions of Bazel should keep
working with the new code until 2018-01-31, i.e.
for the next ~6 months. After that I'll remove the
old flags and the PathListConverter.
*** Reason for rollback ***
Fixed broken integration with release Blaze.
*** Original change description ***
Automated rollback of commit 5752463ece84ebb4fb074888cba57412ab8d86b3.
*** Reason for rollback ***
Broke too many targets.
***
RELNOTES: none
PiperOrigin-RevId: 162587548
|
|
|
|
| |
PiperOrigin-RevId: 162408227
|
|
|
|
|
|
|
|
|
|
|
|
| |
Restrict visibility of the Skylark constructor to the package, which
is still necessary because ConfigSkylarkCommon exposes it, otherwise
it would be private.
It matters less now that these providers are type-safe, but it's
still slightly cleaner.
RELNOTES: None.
PiperOrigin-RevId: 162367759
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Caused memory regression.
*** Original change description ***
CustomCommandLine.Builder: clean up its interface
In this commit:
- remove unused methods and classes
- turn CustomCommandLine.ArgvFragment into an
interface
- remove the
CustomCommandLine.TreeFileArtifactArgvFragment
abstract class; it only had one remaining
subclass
- add @Nullable annotations where nulls are fine
- add Precondition checks for non-nullable args
- simplify the interface by removing add* methods
that can be composed of other add* methods; this
makes it easier to see...
***
RELNOTES: none
PiperOrigin-RevId: 162320031
|
|
|
|
|
|
|
| |
The option filters proto dependency can be removed from the OptionsParser. This is in response to option parser users that want to avoid the bazel-internal proto file in their dependencies.
RELNOTES: None.
PiperOrigin-RevId: 162249778
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Dynamically Configured Resource Filtering change 4/6
Add dynamic transitions for controlling resource filtering. These transitions
will be called in the next few reviews.
Specifically, one transition turns dynamically configured resource filtering
on, and updates the dynamically passed resource filter to contain the current
target's resource filtering settings. The other turns off dynamic filtering,
including removing all filtering settings and transitioning to
FilterBehavior.FILTER_IN_ANALYSIS.
We apply the appropriate transition in getTopLevelPatchTransition: if dynamic
filtering was not requested, no transition is made; if dynamic filtering was
requested but would not be appropriate (either because multiple top-level
targets would mean dynamic configuration would probably split the build graph,
or because no resource filters were provided so dynamically propagating them
would be a waste of time) we transition to static filtering, and, otherwise, we
continue to use dynamic filtering but preserve the current target's filtering
settings so they can be passed to its dependencies.
These transitions will actually be called in the next review.
RELNOTES: none
PiperOrigin-RevId: 162209042
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 162205838
|
|
|
|
| |
PiperOrigin-RevId: 162194755
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit:
- deprecates PathListConverter
- removes ExistingPathListConverter because it was
not used in production, only tests
- deprecated List<Path> type flags that use
PathListConverter
- introduces new List<Path> type flags next to the
deprecated ones that use @Options.allowMultiple
and convert with PathConverter; the new and old
lists are concatenated, yielding the flag value
PathListConverter and all of its occurrences
should be removed after 2018-01-31 (about 6 months
from now, which is a safe enough timeframe for
everyone to upgrade Bazel so it uses the new-style
flags).
Reason for deprecation is that colon-separated
path lists don't work on Windows because paths
have colons in them.
Since the Android BusyBox is not intended to be
executed by users but by Bazel only, there's no
release notes necessary.
See https://github.com/bazelbuild/bazel/issues/3264
RELNOTES: none
PiperOrigin-RevId: 162193998
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Dynamically Configured Resource Filtering change 3/6
Resource filtering behaves somewhat differently when dynamically configured.
Resources obtained from dependencies will have already been filtered and do not
need to be filtered again. Resources that were filtered out do not need to be
tracked since resource processing will never receive a reference to them
anyway.
Also, to make builds where ResourceFilter is dynamically configured, add equals
and hashCode methods (otherwise, the configuration code throws a
NullPointerException) and a global output prefix (otherwise, conflicts can
occur).
To ensure that the global output prefix (and the ResourceFilter object itself)
is the same regardless of the ordering of filters in the android_binary, build
the filters into a set, not a list, and sort them as part of creating the
ResourceFilter object. This way, for example, objects built with filters
"en,fr", "fr,en", and "en,en,fr" will all end up equal.
Finally, rename the getFilteredResources method to better reflect its new role,
and improve the isPrefiltering method to not try to prefilter if there are no
filters.
Add tests for all of this, and helper methods for all of those tests, including,
most notably, a makeResourceFilter method that instantiates a ResourceFilter
similarly to how it is actually created outside of tests (rather than directly
calling the constructor) and a fake implementation of AttributeMap to support
this.
RELNOTES: none
PiperOrigin-RevId: 162099178
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Dynamically Configured Resource Filtering change 2/6
We need to filter LocalResourceContainer for dynamically configured resource
filtering - it is an input to resource parsing for android_library targets.
Filter it appropriately.
Once it's being filtered, use it to build a filtered ResourceContainer, rather
than building an unfiltered ResourceContainer and then filtering it. This means
filtering gets done sooner, saving a bit of time and work, and isn't done
twice, which would just be bad.
To do this, we pass an unbuilt ResourceContainer.Builder into createApk, rather than a fully built one. Move the setAssetsAndResourcesFrom and setManifest calls to that builder, called every time, into createApk so that we only call them once.
Also, initialize the ResourceFilter object when it's needed, rather than pass
it down from the binary or library. This makes the code a bit cleaner, plus
means library targets will pick up the dynamically configured information once
that exists.
RELNOTES: none
PiperOrigin-RevId: 162007885
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In this commit:
- remove unused methods and classes
- turn CustomCommandLine.ArgvFragment into an
interface
- remove the
CustomCommandLine.TreeFileArtifactArgvFragment
abstract class; it only had one remaining
subclass
- add @Nullable annotations where nulls are fine
- add Precondition checks for non-nullable args
- simplify the interface by removing add* methods
that can be composed of other add* methods; this
makes it easier to see what the callers do with
the Builder
- remove add* methods that add a single argument
followed by a list of other elements (or a
joined string of them); these had a bug in that
they didn't check if the collection was empty
(only that it was not null), and if it was empty
then the single argument was still added though
it was not followed by any value
- fix call sites of add* methods where we
previously could have added a flag with an empty
collection
- audit every affected call site
RELNOTES: none
PiperOrigin-RevId: 161957521
|
|
|
|
| |
PiperOrigin-RevId: 161910195
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 161831232
|
|
|
|
|
|
|
|
|
|
|
|
| |
(Almost) all native declared providers are accessed as such and not as
native non-declared providers (inheritors of TransitiveInfoCollaction).
There are still three providers that use
TransitiveInfoCollection.WithLegacySkylarkName mechanism, I'll address
them in the follow-up CL.
RELNOTES: None.
PiperOrigin-RevId: 161655315
|
|
|
|
|
|
|
|
| |
The method is moved to FeaturePolicyConfiguration.java so that it can be used
by unrelated parts of the code using feature policies for whitelisting.
RELNOTES:none
PiperOrigin-RevId: 161648169
|
|
|
|
|
| |
RELNOTES: None
PiperOrigin-RevId: 161588977
|
|
|
|
| |
PiperOrigin-RevId: 161563102
|
|
|
|
|
|
|
|
|
|
| |
OptionMetadataTags.
These are similar, no need to have both fields. Removing the "DOCUMENTED" default, the absence of UNDOCUMENTED will be used instead.
Since requiring a documentation category for undocumented options doesn't make sense, list that as one of the OptionDocumentationCategories, but list HIDDEN and INTERNAL as part of OptionMetadata. These options should list UNDOCUMENTED as their category.
PiperOrigin-RevId: 161515674
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 161395570
|