| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Breaks all projects using Bazel, see https://ci.bazel.io
*** Original change description ***
Deprecated and removed HOST_CFG and DATA_CFG global variables.
--
MOS_MIGRATED_REVID=133005398
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=132976702
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=132464865
|
|
|
|
|
|
|
|
|
| |
string that points to a remote repository.
Fixes #1442.
--
MOS_MIGRATED_REVID=132320130
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Replaced late bound attributes with computed default attributes) with two bug fixes:
1. Unlike SkylarkComputedDefault, SkylarkComputedDefaultTemplate did not sort the names of its attribute dependencies. Consequently, lookup operations failed when callback functions in bzl files specified the names of their required attributes in a non-alphabetical order since the order of the key tuples was different (e.g. [1, 2] vs [2, 1]).
It would be less error prone to always sort the dependencies in createDependencyAssignmentTuple(), but this would impact performance.
2. SkylarkCallbackFunction ignores the legacy "cfg" parameter in callback functions. This special case should be deleted once all cfg parameters have been removed from the depot.
--
MOS_MIGRATED_REVID=132235927
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=132058819
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Motivation:
Compared to computed default attributes, late bound attributes are
evaluated in a later phase (analysis instead of loading phase). While
both mechanisms provide access to other attributes of a rule, only
late bound attributes additionally provide access to the build
configuration.
However, late bound attributes could be used to return new labels that
have not been loaded before. Since this happens in the analysis phase,
it can break one of Blaze's underlying principles, thus introducing a
serious correctness bug.
We decided to replace late bound attributes in Skylark with computed
default attributes since this moves the evaluation of values into the
loading phase, thus fixing this bug. Moreover, none of the existing
users of this mechanism required access to the build configuration,
which means that the user impact of this change is minimal.
Implementation details:
Unlike attributes of non-Skylark rules, however, Skylark computed
defaults need to be able to depend on more than two configurable
attributes since all attributes of Skylark rules are configurable by
default. This has two implications:
1. Unlike "normal" Skylark attributes, Skylark computed defaults need
to know about the existence of other attributes in order to declare a
dependency on them. This CL takes advantage of work previously done to
require parameter names to be the names of attributes used by the
computed default function.
2. Since Bazel computes the combinations of all possible attribute
values, a Skylark rule with a computed default that depends on a lot
of configurable attributes could crash Bazel. Consequently, this CL
also introduces an upper bound (64) for the number of valid
combinations.
Caveats:
1. Getting the depended-on attributes' names from function paramters is
mildly surprising.
Alternatives:
The best solution would be to keep SkylarkLateBound, but restrict it in
a way that it can only return already loaded labels. This is not
possible right now.
--
MOS_MIGRATED_REVID=131967238
|
|
|
|
|
|
|
| |
function.
--
MOS_MIGRATED_REVID=129726780
|
|
|
|
|
|
|
|
|
|
| |
With the prereq work behind this, this is surprisingly straightforward. The main change
is to eliminate BuildConfiguration.SplittableTransitionApplier, make both DynamicTransitionApplier and StaticTransitionApplier split-aware, and add awareness of this to ConfiguredTargetFunction.trimConfigurations.
Latebound splits will follow next.
--
MOS_MIGRATED_REVID=129480309
|
|
|
|
|
|
|
| |
This in preparation to DeclaredProviders implementation.
--
MOS_MIGRATED_REVID=129420617
|
|
|
|
|
|
|
|
|
|
|
| |
This change was motivated by a need to write pure Skylark rules that expose their own objc providers so they can be used as deps to other libraries/application targets (e.g., SceneKit/SpriteKit compiled resources, []) without having to whitelist them and wait for a Blaze release.
This CL fixes what seems to be a bug in validateRuleDependency, where the behavior in the doc comment implies that it will accept a whitelisted rule name *or* a list of mandatory providers, but as implemented today it seems to require the rule to be whitelisted even if the mandatory native providers matched.
RELNOTES: objc_* rules can now depend on any target that returns an "objc" provider.
--
MOS_MIGRATED_REVID=128835096
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
for general readability.
Major changes:
- Remove the intermediate Attribute -> LabelAndConfiguration multimap (computed in resolveAttributes). Instead, feed discovered values directly into the final Attribute -> Dependency map via a new RuleResolver interface.
- Remove all references to LabelAndConfiguration. The configuration is always the owning rule's configuration except for two special cases: late-bound attributes with splits and late-bound attributes with LateBoundDefault.useHostConfiguration. The original interface made this very unclear and required a lot of awkward and sometimes incorrect logic. The new interface only involves configurations for the cases that actually need them.
- Remove an ugly hack caused by BuildConfiguration.evaluateTransition mixing poorly with LateBoundDefault.useHostConfiguration (https://github.com/bazelbuild/bazel/blo[]e172693c27f3efc95ed163e43a9f0a7a6fb4017/src/main/java/com/google/devtools/build/lib/analysis/DependencyResolver.java#L488).
- Remove a hack that applies split transitions twice because of BuildConfiguration.evaluateTransition mixing poorly with late-bound split attributes (https://github.com/bazelbuild/bazel/blo[]e172693c27f3efc95ed163e43a9f0a7a6fb4017/src/main/java/com/google/devtools/build/lib/analysis/DependencyResolver.java#L319). This happens to be innocent now but won't be when nested splits are possible.
- Solidifies the API contract for Attribute.LateBoundDefault.useHostConfiguration.
- Applies clearer naming and more consistent ordering to method parameters.
- Better documentation.
This is all also prep work for dynamic split transitions.
tl;dr: late-bound attributes are legitimately special. Treat them that way to make the rest of DependencyResolver cleaner and hack-free.
--
MOS_MIGRATED_REVID=128582618
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=128510907
|
|
|
|
|
|
|
| |
Also clarify that the returned list is immutable.
--
MOS_MIGRATED_REVID=128482720
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127808009
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=126620866
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=126081020
|
|
|
|
|
|
|
|
|
|
|
|
| |
validation and conversion for different types of attributes.
Two reasons for this change:
1. These tasks shouldn't be the business of SkylarkRuleClassFunctions which previously handled such things.
2. With unknown commit we have to take care of three different types of attributes (normal vs. late-bound vs. computed default), which means that a boolean argument is no longer sufficient.
--
MOS_MIGRATED_REVID=122136679
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There are no syntactic changes within Skylark; the only difference is that aspects may have non-implicit attributes, so long as they have type 'string' and use the 'values' restriction. Such aspects may only be requested by rules which have attributes with types and names matching the non-implicit, non-defaulted attributes of the aspect.
This is not yet a complete match for native AspectParameters functionality since implicit attributes cannot yet be affected by attribute values, but that will be added later.
Implicit aspects are still required to have default values. Non-implicit aspect attributes are considered "required" unless they have a default value. An error will occur if they are applied to a rule that does not "cover" all required attributes by having attributes of matching name and type. While non-implicit aspect attributes with a default are not required, they will still take on the value of a rule attribute with the same name and type if it is present.
Aspects with non-implicit, non-defaulted ("required") attributes cannot be requested on the command line, only by a rule.
RELNOTES: Expose parameterized aspects to Skylark.
--
MOS_MIGRATED_REVID=121711715
|
|
|
|
|
|
|
| |
and use it to validate that :java_toolchain has a JavaToolchainProvider.
--
MOS_MIGRATED_REVID=121396726
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
NativeAspectClass.
This a large refactoring of the aspects, currently we have the following:
- AspectClasses: The interface AspectClass is a implemented by either
SkylarkAspectClass or NativeAspectClass<NativeAspectFactory>.
They are wrappers for the AspectFactories and they hold the information about
the Class<> of the factory.
- AspectFactories (FooAspect.java): Represented by the interfaces
ConfiguredAspectFactory and NativeAspectFactory, also by
the interface ConfiguredNativeAspectFactory which is the union of the two
aforementioned interfaces.
All aspects implement ConfiguredNativeAspectFactory except Skylark aspects
which implement only ConfiguredAspectFactory.
After this CL the distinction between NativeAspectFactories and NativeAspectClasses
dissappear, namely aspect that extends NativeAspectClass is considered native
and if it implements ConfiguredAspectFactory it is configured.
Therefore the interfaces NativeAspectFactory and ConfiguredNativeAspectFactory
both disappear.
With this refactoring the aspectFactoryMap in the ConfiguredRuleClassProvider
changes its type from (String -> Class<? extends NativeAspectClass>)
to (String -> NativeAspectClass) which means it is now able to have an instance
of the aspect instead of its Class only.
By doing this, it is now possible to pass parameters when creating an
aspect in the ConfiguredRuleClassProvider.
--
MOS_MIGRATED_REVID=120819647
|
|
|
|
|
|
|
| |
attribute based on the Rule itself (the transition may thus be determined based on the values of other attributes of the rule)
--
MOS_MIGRATED_REVID=120275649
|
|
|
|
|
|
|
| |
Its old name was confusing because resolve() and getDefault() do radically different things: getDefault() returns a good enough lie for when BuildConfiguration is not available, and resolve() resolves the dependency when we do have a BuildConfiguration.
--
MOS_MIGRATED_REVID=120212630
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=119162307
|
|
|
|
|
|
|
|
|
|
| |
Aspect becomes a triple (AspectClass, AspectDefinition,
AspectParameters) and loses its equals() method.
After this CL, SkylarkAspectClass.getDefintion still exists and is
deprecated.
--
MOS_MIGRATED_REVID=119159653
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=118699141
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=117202268
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=116980421
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Using mandatoryProvidersList to validate python rules' dependency.
Added a SkylarkProvider named 'py' which is a SkylarkClassObject in Java and a
struct in Skylark. Native python rule and Skylark python rule should have this provider
so that they can depend on each other.
RELNOTES[NEW]: Native python rule can depend on skylark rule as long as skylark
rule provides 'py' provider.
--
MOS_MIGRATED_REVID=116241504
|
|
|
|
|
|
|
|
| |
"mandatoryProvidersList" is a list of sets of providers. For any rule, if it provides
all the providers from one of those sets, we consider the dependency valid.
--
MOS_MIGRATED_REVID=115221394
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
https://github.com/bazelbuild/bazel/blo[]fbbd6a32b95ba746f09dae1eaeaccf675cd5b3/src/main/java/com/google/devtools/build/lib/packages/Attribute.java#L1045
This allows the default value computation for latebound attributes to consider
the values of configurable attributes. This is most directly useful for user-definable Skylark defaults, which have full access to the values of all non-latebound attributes.
Without this change, this kind of scenario crashes Bazel. For example:
------------------
select_rules.bzl:
------------------
def _impl(ctx):
ctx.file_action(
output=ctx.outputs.out_file,
content=ctx.attr.string_value,
)
return struct()
# Bug does not manifest without using this as a default.
def _derived_value(attrs, _):
return Label("//some:dep")
selector_rule = rule(
implementation=_impl,
attrs={
"string_value": attr.string(default=""),
"out_file": attr.output(),
"_derived": attr.label(default=_derived_value),
},
output_to_genfiles=True,
)
def selector_macro(name, out_file="", string_value=""):
# This will fail with selectors.
selector_rule(
name="%s_skylark" % name,
string_value=string_value,
out_file=out_file + ".skylark",
)
# This does not.
native.genrule(
name="%s_genrule" % name,
cmd="echo '" + string_value + "' > $@",
outs=[out_file + ".genrule"],
)
native.filegroup(
name=name,
srcs=[":%s_genrule" % name, "%s_skylark" % name],
)
------------------
BUILD.bzl:
------------------
config_setting(
name = "selector",
values = {"compilation_mode": "opt"},
)
selector_macro(
name = "this_rule",
string_value = """soup? """ + select({
":selector": "no, thank you.",
"//conditions:default": "yes, please!!",
}),
out_file = "this_rule.txt",
)
--
MOS_MIGRATED_REVID=114326474
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Skylark into their own package. This allows, e.g., classes in the syntax package to access classes in the cmdline package without creating circular dependencies.
While we're here:
- Removed a couple of unused BUILD deps flagged in [].
- Updated SkylarkRuleImplementationFunctionsTest to remove non-ASCII characters and
clarify the intent of the test.
--
MOS_MIGRATED_REVID=110360763
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=110356954
|
|
|
|
|
|
|
| |
Reduces garbage.
--
MOS_MIGRATED_REVID=109914243
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=108777120
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=108243881
|
|
|
|
| |
MOS_MIGRATED_REVID=107604619
|
|
|
|
|
|
|
|
| |
Aspect => ConfiguredAspect
AspectWithParameters => Aspect
--
MOS_MIGRATED_REVID=107375211
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=106848269
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=106694515
|
|
|
|
|
|
|
|
|
|
|
| |
attributes that would otherwise be skipped by
default policy.
This is the simplistic start to a user-controllable
enforcement policy API.
--
MOS_MIGRATED_REVID=106530210
|
|
|
|
|
|
|
| |
For native aspects, AspectClass is a facade for Class<AspectFactory<...>>.
--
MOS_MIGRATED_REVID=105986390
|
|
|
|
|
|
|
|
|
|
|
| |
The headers were modified with
`find . -type f -exec 'sed' '-Ei' 's|Copyright 201([45]) Google|Copyright 201\1 The Bazel Authors|' '{}' ';'`
And manual edit for not Google owned copyright. Because of the nature of ijar, I did not modified the header of file owned by Alan Donovan.
The list of authors were extracted from the git log. It is missing older Google contributors that can be added on-demand.
--
MOS_MIGRATED_REVID=103938715
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
to Attribute.skipPrereqValidatorCheck.
This is to disambiguate the concept of "constraints" and keep
the word consciously focused on Bazel's *new* constraint system.
The changed methods refer to checks done by PrerequisiteValidator,
which is basically an adhoc version of the "old" system (e.g.
checking visibility)
--
MOS_MIGRATED_REVID=103872412
|
|
|
|
|
|
|
|
|
|
| |
- Label parsing can be simplified
- lib.syntax is only contains the code for Skylark and is reasonably independent from the problem domain of building things
This change is mostly only changes to imports declarations. The rest is reversing the dependency between :cmdline and :syntax and moving a tiny amount of code between Printer and FilesetEntry and the addition of SkylarkPrintableValue that I couldn't be bothered to separate out into its own change.
--
MOS_MIGRATED_REVID=103527877
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=103374106
|
|
|
|
|
|
|
| |
BaseFunction.
--
MOS_MIGRATED_REVID=102988766
|
|
|
|
|
|
|
|
|
| |
The lipo dependency is artificial; it's an artifact of how LIPO is implemented
in Bazel. Running these checks doesn't make sense; they unnecessarily
disallow perfectly valid scenarios.
--
MOS_MIGRATED_REVID=102550286
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=102126786
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a big change, so let me walk you through the key pieces:
1) This cl provides an alternative mechanism for creating configurations and doing configuration transitions that's "dynamic" in that the configurations can be created on the fly and the transitions are arbitrary mappings from BuildOptions --> BuildOptions that can also be created on the fly. It also integrates this with ConfiguredTargetFunction, so the configured target graph automatically uses this framework.
2) It does *not* replace old-style (which we'll call "static") configurations. For a number of important reasons: It's not yet at feature parity (particularly: no LIPO). It's not remotely tested across real projects enough to have confidence that it's battle-ready. It doesn't yet handle certain "special" functions like BuildConfiguration.prepareToBuild() and BuildConfiguration.getRoots(). It still relies on the old static transition logic to determine what transitions to apply (eventually we'll distribute that logic out, but that's too much for a single cl). We need the option to toggle it on and off until we have enough confidence in it. So with this cl, builds can be done in either mode.
3) The new flag --experimental_dynamic_configs toggles use of dynamic configurations.
4) Dynamic configurations are created with the Skyframe function BuildConfigurationFunction (this was already created in an earlier change). This consumes a BuildOptions and a set of configuration fragments to produce a BuildConfiguration.
5) Dynamic transitions are instances of the new class PatchTransition, which simply maps an input BuildOptions to an output BuildOptions.
6) Since static and dynamic configurations have to co-exist (for now), this cl tries hard to keep today's transition logic defined in a single place (vs. forking a dedicated copy for each configuration style). This is done via the new interface BuildConfiguration.TransitionApplier. BuildConfiguration.evaluateTransition is modified to feed its transition logic into TransitionApplier's common API. Both dynamic and static configurations have their own implementations that "do the right thing" with the results.
7) The transition applier for dynamic configurations basically stores the Transition, then allows DependencyResolver (which calls BuildConfiguration.evaluateTransition) to return Dependency instances containing that Transition (vs. a BuildConfiguration, which they traditionally contain).
7.5) An earlier variation of the dynamic transition applier retained BuildOptions (e.g. when it got a Transition it immediately applied it to get its output BuildOptions, then stored that). This had the advantage of making composing of transitions easier, especially within BuildConfiguration.evaluateTransition (which can theoretically apply multiple transitions to the input configuration). But it turns out that applying transitions has a cost, and it's simply more performant to pass them through until they're really needed.
8) In dynamic configuration mode, after ConfiguredTargetFunction gets its deps (e.g. an <Attribute, Dependency> multimap), it "trims" the configurations for its dependencies by a) only including config fragments required by the deps' subtrees and b) applying the transitions that came from 7). This all happens in the new method ConfiguredTargetFunction.trimConfigurations.
9) trimConfigurations is heavily performance-optimized based on a lot of experience running this through a large project within Google. As it turns out, the cost of host transitions can be atrocious (because there are a lot of them). Also, BuildOptions.clone() is expensive. And just creating BuildConfiguration SkyKeys also has a cost (largely because of BuildOptions cloning), so that shouldn't be done except when really necessary. My experience with this convinced me it's worth making this method complicated for the sake of making it fast. Since it basically visits every edge in the configured target graph (at least), it really needs to be slick.
10) Since host transitions in particular are problematic w.r.t. speed, I compute the host *once* in ConfigurationCollectionFunction.getHostConfiguration() and expose that reference to ConfiguredTargetFunction and other Skyframe functions. This limits recomputation to just when the fragments are trimmed.
11) Since options cloning is expensive, I'm doing something scary: exposing a BuildConfiguration.getOptions() method that returns a direct reference. Since BuildOptions is mutable, this is dangerous in the wrong hands. I can experiment with going back to cloning (with the caching of host transitions it may not matter as much), but we may ultimately have to put up with this risk for the sake of performant analysis time. What would be *really* awesome would be to make BuildOptions immutable. But that's not going to happen in this cl.
So in short, the key abstractions in this cl are:
- PatchTransition
- BuildConfiguration.TransitionApplier
- ConfiguredTargetFunction.trimConfigurations
The current implementation imposes no analysis time penalty
--
MOS_MIGRATED_REVID=101474620
|