| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
|
|
|
| |
Classes of the same EvalUtils.getSkylarkType are once again the same for the
type inferencer.
Also, for debugging purposes, only print SkylarkClassObject as struct, not all classes that implement ClassObject yet are considered disjoint by the type engine.
--
MOS_MIGRATED_REVID=87933890
|
|
|
|
|
|
|
| |
This fixes flakiness on underpowered machines.
--
MOS_MIGRATED_REVID=87824341
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87698104
|
|
|
|
|
|
|
| |
Change-Id: I83a280d53272972bace811b1493825b5dc48a4d2
--
MOS_MIGRATED_REVID=87608375
|
|
|
|
|
|
|
|
| |
methods available doesn't seem to carry its weight, and this makes it easier
for us to migrate to JUnit 4.
--
MOS_MIGRATED_REVID=87597162
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87596401
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87595869
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87594852
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87535290
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87513766
|
|
|
|
|
|
|
|
|
|
|
| |
Fix race condition in SkylarkType.of(), whereby .equals() but not same (==) types were created, by making Simple.of() synchronized.
Also make SkylarkType#includes() more robust by using .equals() instead of == (it's a little bit slower in the case of Simple types once fixed, but also works on complex types that don't hash-cons their values).
Also, distinguish SkylarkList (printed as list) and java.util.List (printed as List) and similarly for tuple vs Tuple, when printing types in debugging messages.
--
MOS_MIGRATED_REVID=87490176
|
|
|
|
|
|
|
| |
This is needed for taking the runfiles prefix from the WORKSPACE file instead of hardcoding it.
--
MOS_MIGRATED_REVID=87347883
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87345558
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87342725
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Refactor SkylarkType, notably adding Union types and runtime typechecks.
These are pre-requisites to unifying all Skylark function calls to use the same
code path, that will check types (like SkylarkFunction currently does) as well
handle a more complete Python calling convention (like MixedModeFunction is
almost but not quite able to do).
A SkylarkType can be either
* a Simple type corresponding to a Java class, with special types TOP and BOTTOM
corresponding respectively to Object and EmptyType (similar to Void).
* a Combination of a generic class (LIST, MAP or SET) and an argument type
* a Union of a finite number of types
* a FunctionType associated with a name and a returnType (with ugly
validation-time side-effects that we should probably move out of the type)
Unions are necessary because:
1- the type of some builtin function arguments are actually the Union of some
type and a function type for a callback that computes the actual value.
2- the type of some builtin function arguments declared as "List" is actually
the Union of java List and SkylarkList.
3- instead of adding lots of special cases at the point of argument validation,
it's cleaner and more generally useful to have explicit Union types, that can
then be displayed to users for documentation or debugging purposes.
Validation-time "type inference" is re-expressed simply as type intersection.
None is treated specially as inferred into TOP instead of NoneType.
Combination types are printed as genericType of argTypes, e.g. "dict of ints". "type" and "generic1" become "genericType" and "argType". In SkylarkList and SkylarkNestedSet, genericType becomes contentType.
--
MOS_MIGRATED_REVID=87340881
|
|
|
|
|
|
|
|
|
|
|
| |
A recent change resulted in TopLevelArtifactHelper.getTopLevelArtifacts()
returning an empty result when passed TopLevelArtifactContext.DEFAULT.
This only happens in tests and wasn't caught because tests were never
fully checking the results. Update MultiGroupForgeResourceAccountantTest
to better consider/expose this failure and fixed.
--
MOS_MIGRATED_REVID=87254923
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87248208
|
|
|
|
|
|
|
| |
TestCase for it.
--
MOS_MIGRATED_REVID=87246613
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87244284
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Users have asked for ways to control the concurrency level of their
local tests. They can do it right now using the --local_resources
option, but that is unintuitive and affects the parallelism of
non-test actions.
This option changes the kind of resources obtained for local tests. If
set, the CPU, RAM, and IO dimensions for local tests will not be used,
and a new localTestCount dimension will be used, where the capacity is
equal to the option's value, and each local test consumes one unit.
--
MOS_MIGRATED_REVID=87177698
|
|
|
|
|
|
|
|
|
|
|
|
| |
tree, rewrite most uses of assertPresence. This is a small step towards
migrating the tests to JUnit 4. Depending on inheritance for the assert*
methods is not recommended, and the (old) Asserts class is deprecated in
JUnit 4.
Sorry, there was no good way to do this piecemeal.
--
MOS_MIGRATED_REVID=87154212
|
|
|
|
|
|
|
|
| |
the tests under analysis/actions. They don't run yet, because the mock client
setup is still missing.
--
MOS_MIGRATED_REVID=87149625
|
|
|
|
|
|
|
| |
load.
--
MOS_MIGRATED_REVID=87143293
|
|
|
|
|
|
|
| |
are unused.
--
MOS_MIGRATED_REVID=87139146
|
|
|
|
|
|
|
| |
instead of nanoTime. As nanoTime should only be used to compute time differences. To avoid having to pass two long values representing time (And calling twice to Clock.xxx methods per action executed) we compute the wall time by passing a clock and computing the difference from the start nano time.
--
MOS_MIGRATED_REVID=87102963
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87066171
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=87036991
|
|
|
|
|
|
|
|
|
|
| |
Most current invocations of the ResourceSet constructor shouldn't
care about any new dimensions we add to restrict test execution. This
change allows us to make those changes, keep the ResourceSet ctor
symmetrical, and avoid contaminating those ctor call sites with noise.
--
MOS_MIGRATED_REVID=86826112
|
|
|
|
|
|
|
|
|
|
|
|
| |
While attempting to clean up the Bazel examples, I noticed that the example
tests don't actually work because they expect a third_party directory. We could
link third_party/ from the top-level bazel directory, but then we'd have
crossing symlinks (base_workspace/tools->tools and
third_party->base_workspace/third_party). Linking everything in one direction
seemed like a better option, but alternative suggestions are welcome.
--
MOS_MIGRATED_REVID=86703332
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Also includes the following changes:
Skylark: tests copy the Skylark files into the same directory. Load statements are replaced with relative paths within bzl files and with absolute paths in tests.
--
Refactor Objc abstract rule classes.
This new arrangement groups attributes by logical purpose and clarifies
rule composition. For example objc_binary inherits "linking" whereas an
objc_library inherits "linkable" but both inherit "compiling".
The only functional change in this CL is to remove the "data" attribute
from all rules as it is not used in the depot and doesn't make sense
when we also allow specifying resources in any number of other ways.
--
MOS_MIGRATED_REVID=86676190
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=86431291
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=86430394
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=86427416
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=86171408
|
|
|
|
|
|
|
| |
BuildRuleBuilder to define dependencies between rules.
--
MOS_MIGRATED_REVID=85887994
|
|
--
MOE_MIGRATED_REVID=85702957
|