| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
|
|
|
|
| |
* FILE_SYMLINK* codecs need ImmutableList so this adds an ImmutableList runtime codec.
* Adds SkyFunction names where codecs exist for their SkyKeys.
* Adds a few @AutoCodec singleton codecs.
We cannot yet disable the ImmutableList marshaller because currently the
runtime ImmutableList codec requires runtime codecs for the contained elements, which doesn't hold everywhere (particularly for protos).
PiperOrigin-RevId: 187224736
|
|
|
|
| |
PiperOrigin-RevId: 187222454
|
|
|
|
| |
PiperOrigin-RevId: 187218309
|
|
|
|
| |
PiperOrigin-RevId: 187218260
|
|
|
|
| |
PiperOrigin-RevId: 187212799
|
|
|
|
| |
PiperOrigin-RevId: 187212436
|
|
|
|
| |
PiperOrigin-RevId: 187209783
|
|
|
|
|
|
|
|
|
| |
This is only used by the static method of the same name, which is only used by RuleFactory.
This method is showing up in profiles as creating more garbage than I'm happy with, but not
enough to make me drop everything and fix it now. Instead make it private so folks are less
inclined to use it later...
PiperOrigin-RevId: 187208467
|
|
|
|
| |
PiperOrigin-RevId: 187199625
|
|
|
|
|
|
|
|
| |
Avoid throwing an execption if we don't have to, stack traces are expensive to
fill in.
RELNOTES: None
PiperOrigin-RevId: 187199392
|
| |
|
|
|
|
|
|
|
|
|
| |
case where a pattern like '*.blah' doesn't match all the dirents.
Before the recent Path.java rewrite, the unconditional allocation here was free/amortized (in the sense that if Blaze ever had to consider the path in question, then we'd be doing the Path allocation anyway and caching the object). But now it's potentially GC churny, especially in the case where a single BUILD file has multiple glob expressions.
RELNOTES: None
PiperOrigin-RevId: 187185573
|
|
|
|
|
|
| |
This flag changes the behavior of objc_library module map propagation so that module maps are only propagated to direct dependents, not transitive dependents. swift_library targets that import Objective-C code must then list those dependencies directly in its deps instead of depending on them being transitively present.
PiperOrigin-RevId: 187184692
|
|
|
|
|
|
|
|
| |
This requires codecs for CustomCommandLine and a Marshaller for Charset.
Added AutoCodec functionality to generic classes like Iterable<T>.
PiperOrigin-RevId: 187182889
|
|
|
|
|
|
|
| |
Progress on #4544.
RELNOTES: None.
PiperOrigin-RevId: 187179454
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Make the --experimental_repository_cache option no longer experimental and enable
the cache by default. The default location is in the a separate directory under the
install base (under .cache/bazel user's home directory); this means it is shared
between workspaces und not between different versions of bazel. We plan to also
share it between different bazel versions in the future.
Fixes #4676.
RELNOTES:
repository_cache is no longer experimental and enabled by default.
Change-Id: I8499c2d1811e0fe8d830796e07cd6f8bc75e48ba
PiperOrigin-RevId: 187172766
|
|
|
|
|
|
|
|
|
|
|
| |
...so that it can use that path to compute other directories in the
output user base, in particular the default location for caches.
The first cache we will add is the hash-index cache for downloads
of external archives, but a spawn cache might be added later in the
output user base as well.
Change-Id: I24b1c33235c8f76ec008ecb1789163de2b2a45be
PiperOrigin-RevId: 187164275
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
unknown commit fixes the only related broken target showing up in a TGP after
this change. It seems that this provider was only ever used through
java_wrap_cc with java_library dependencies which themselves had cc_library
dependencies, i.e. java_wrap_cc -> java_library -> cc_library. This dependency
is not needed, the cc_library can be added directly to java_wrap_cc to get the
same behavior and this is the way it should be done instead of relying on
transitive dependencies.
The reason for making this change right now is that with the Skylark C++ API, I
will be adding attr("deps").mandatoryProviders(CcCompilationInfo,
CcLinkParamsInfo) to C++ rules so that they can take any Skylark rule and not
having to explicitly hard-code the name of those Skylark rules in Blaze.
However, Java meets this criteria and the change would allow C++ rules to
depend on Java. This is not something that was supported before and is not
something we plan to support.
RELNOTES:none
PiperOrigin-RevId: 187158176
|
|
|
|
| |
PiperOrigin-RevId: 187113491
|
|
|
|
|
|
|
| |
will be used for the dependency checker later. Refactoring it now just avoids future potential merge conflicts.
RELNOTES: None
PiperOrigin-RevId: 187103807
|
|
|
|
|
|
| |
come from Skylark, so we shouldn't crash. As a safety measure, subclasses of ActionLookupValue are now responsible for detecting action conflicts themselves.
PiperOrigin-RevId: 187095271
|
|
|
|
|
|
|
| |
:packages and checked-in gRPC jars.
RELNOTES: None
PiperOrigin-RevId: 187088590
|
|
|
|
|
|
|
|
|
|
| |
I'm assuming most of the time folks are only assigning one variable, which
doesn't require an ImmutableSet.Builder and the garbage it comes with.
Also took the liberty of removing some unused InterruptedException
declarations.
PiperOrigin-RevId: 187068400
|
|
|
|
|
|
| |
that has a single static INSTANCE field pointing back to the target field, so that serialization can grab it.
PiperOrigin-RevId: 187065629
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Only check/update type if the types of elements that we're adding changes. In
particular, the "is DICT/LIST" check can get expensive when run on every single
inserted element.
According to a hacky unit test (create 10M element SkylarkNestedSet 10x over) this is ~60%
faster (~18s -> ~7s).
This change is intended to be a quick fix, there's surely a more principled way
to make skylark's type checking more efficient...
PiperOrigin-RevId: 187062547
|
|
|
|
| |
PiperOrigin-RevId: 187059719
|
|
|
|
| |
PiperOrigin-RevId: 187057628
|
|
|
|
| |
PiperOrigin-RevId: 187057332
|
|
|
|
| |
PiperOrigin-RevId: 187054284
|
|
|
|
|
|
| |
This field is needed by the Android Studio with Bazel plugin.
PiperOrigin-RevId: 187053906
|
|
|
|
|
|
|
|
|
|
| |
AutoCodec still cannnot handle types with generic elements, e.g.
class Foo<T> {
private T member;
}
PiperOrigin-RevId: 187052487
|
|
|
|
| |
PiperOrigin-RevId: 187048921
|
|
|
|
| |
PiperOrigin-RevId: 187044297
|
|
|
|
| |
PiperOrigin-RevId: 187043046
|
|
|
|
|
|
| |
class in https://github.com/bazelbuild/bazel/commit/14e549cc6596f67ed22c1bac0819f0775e4fb806).
PiperOrigin-RevId: 187042482
|
|
|
|
| |
PiperOrigin-RevId: 187038647
|
|
|
|
|
|
| |
Most of the time there's only one, doesn't justify the iterator overhead.
PiperOrigin-RevId: 187031802
|
|
|
|
| |
PiperOrigin-RevId: 187030267
|
|
|
|
|
|
|
|
|
|
|
| |
This cl introduces a bazel option
(--experimental_fix_linkstamp_inputs_bug_73447914) to control which inputs get
added into C++ linkstamp compile action. When set to true (defaults to false),
all inputs of relevant C++ linking action get added as inputs to the linkstamp
compile action too.
RELNOTES: None.
PiperOrigin-RevId: 187030217
|
|
|
|
| |
PiperOrigin-RevId: 187029559
|
|
|
|
| |
PiperOrigin-RevId: 187029004
|
|
|
|
| |
PiperOrigin-RevId: 187028864
|
|
|
|
| |
PiperOrigin-RevId: 187028110
|
|
|
|
|
|
|
|
|
|
|
| |
- collapsed these into one section
- removed the term "implicit outputs"
- add explanation for what these are first, then how to control them
- update docs for DefaultInfo
- also update docs for test/executable args of rule()
RELNOTES: None
PiperOrigin-RevId: 187026641
|
|
|
|
|
|
|
| |
This is a change from behavior introduced in unknown commit, which was added to preserve determinism even though the ordering of dicts in Python is non-deterministic. Now that we don't call into Python anymore, this is not necessary.
RELNOTES: None (select branches are mutually exclusive, so this is a no-op)
PiperOrigin-RevId: 187022575
|
|
|
|
|
| |
RELNOTES: None
PiperOrigin-RevId: 187018017
|
|
|
|
|
|
|
| |
Unified "Files" and "Output files" sections. Renamed "declared files" to "predeclared files" to avoid confusion with "declare_file()"/"declare_directory()".
RELNOTES: None
PiperOrigin-RevId: 187017607
|
|
|
|
|
|
|
|
|
|
| |
Dynamic environment variables are going away.
This also fixes a crashing bug.
Fixes #3983
RELNOTES: None.
PiperOrigin-RevId: 187008011
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Stamped binaries now have a correct
build-data.properties file with regard to the
"Build time" field.
There's a mismatch in BuildInfo.BUILD_TIMESTAMP
semantics between Bazel and Google-internal Blaze:
Bazel assumes this amount is in milliseconds,
Blaze assumed seconds.
Since there are users depending on this field
already on both sides, it is too late to change
its semantics. So this commit only fixes the
bogus "Build time" field and leaves
BUILD_TIMESTAMP as milliseconds for Bazel and as
seconds for Blaze.
Fixes https://github.com/bazelbuild/bazel/issues/4469
Change-Id: I28282d660244af693ab16443ae1e42fb455f9dde
PiperOrigin-RevId: 187005016
|
|
|
|
|
|
|
|
| |
This is a change from behavior introduced in unknown commit, which was added to preserve determinism even though the ordering of dicts in Python is non-deterministic. Now that we don't call into Python anymore, this is not necessary.
RELNOTES[INC]: The order of dict-valued attributes is now the order in the BUILD file (or in the Skylark dict they were created from) and not lexicographically sorted.
PiperOrigin-RevId: 187003317
|