| Commit message (Collapse) | Author | Age |
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127979748
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
ActionInputFileCache:
Change getDigest() to return the underlying byte[16] owned by each
FileArtifactValue.
Remove throws clause from getInputFromDigest(); this should be an
in-memory operation, and no implementation actually throws.
PerActionFileCache:
Invert mapping from artifact to digest only if needed. Remove interner,
as it was used only for the reverse map keys, not the returned values.
This should be a significant cpu savings as eagerly constructing the
reverse maps was a noticeable hotspot.
--
MOS_MIGRATED_REVID=127972359
|
|
|
|
|
|
|
| |
messages from manifest merging when using manifest_merger="android" on android_binary rules.
--
MOS_MIGRATED_REVID=127962630
|
|
|
|
|
|
|
| |
Fixes #1537.
--
MOS_MIGRATED_REVID=127945452
|
|
|
|
|
|
|
|
|
| |
For instance, for a simple .bzl program that just iterates over a range
of size 200M, the total running time drops from about two minutes to half
a minute.
--
MOS_MIGRATED_REVID=127942625
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The current approach for the py_binary wrapper means that deployed
instances of py_binary targets create child processes. This can be a
problem when you try to kill the py_binary target since signals and
such don't necessarily get passed along to the child process.
Using the exec functionality here should avoid this particular
problem. The overall behaviour should remain the same.
--
Change-Id: I20ef081e1da5b28a8aff8514d713b1c387ae7f7e
Reviewed-on: https://bazel-review.git.corp.google.com/#/c/4070/
MOS_MIGRATED_REVID=127929454
|
|
|
|
| |
MOS_MIGRATED_REVID=127927495
|
|
|
|
|
|
|
|
|
| |
//tools/objc/ios_test.sh.bazel_template. Also remove the iossim substitutions and runfiles in ios_test script register.
[]:
--
MOS_MIGRATED_REVID=127911427
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127895738
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Forgot to do this, so we get compiler errors
since it can't find the R class imported under
the requested custom java package.
Also, avoid writing out the top-level R.class
file if there are no symbols. There is a test
that happens to check for this.
--
MOS_MIGRATED_REVID=127888913
|
|
|
|
|
|
|
| |
exception that caused the crash is an OOM. Too many systems wrap Errors in other exceptions, for instance https://github.com/google/guava/blob/master/guava-gwt/src-super/com/google/common/cache/super/com/google/common/cache/LocalCache.java#L267
--
MOS_MIGRATED_REVID=127845350
|
|
|
|
|
|
|
|
|
| |
Now we can use 'bazel run' to run cpp and java binary!
--
Change-Id: I84b597765dc77127f3fb701d7bd4a81f6dc44660
Reviewed-on: https://bazel-review.googlesource.com/#/c/4054
MOS_MIGRATED_REVID=127834930
|
|
|
|
|
|
|
|
|
|
|
| |
ExecutionInfo contains the tag "support-workers" set to "1" and fallback to non-worker execution if it is not present.
This will eventually allow us to safely automatically decide whether to use workers to execute an action or not.
RELNOTES[INC]: If you maintain a rule that uses persistent workers, you'll have to specify execution_requirements={"supports-workers": 1} in the ctx.action that intends to run a tool with workers. The WorkerSpawnStrategy will alert you with a warning message if you forget to make this change and fallback to non-worker based execution.
--
MOS_MIGRATED_REVID=127822788
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127815874
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127808009
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127807675
|
|
|
|
|
|
|
|
|
| |
Skyframe globbing. This adds a log(n) factor to uses of globs, but getting globs to be returned in a reasonable order that can be emulated by legacy globbing is hard and bug-prone right now, and we must sort anyway if we are merging legacy and Skyframe globs.
Note that this log(n) factor is already present on clean builds with legacy globbing. If we end up seeing performance issues on incremental loading, we can investigate making GlobFunction efficiently return elements in sorted order. (We would still need to sort if merging legacy and Skyframe globs, but that should be a relatively rare occurrence, and can be dealt with by a more efficient merge sort if necessary.)
--
MOS_MIGRATED_REVID=127752554
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127747661
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127735980
|
|
|
|
|
|
|
| |
Fixes #1125.
--
MOS_MIGRATED_REVID=127719941
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Let's launch open-source java_proto_library with strict-deps checking turned on by default.
*** Original change description ***
Unveil Bazel's proto_library and java_proto_library.
--
MOS_MIGRATED_REVID=127716437
|
|
|
|
|
|
|
| |
Instead have callers get it via package.
--
MOS_MIGRATED_REVID=127715494
|
|
|
|
|
|
|
| |
In theory, now we can cancel Ping() and Cancel() RPCs, too, but since we don't tell their UUID anyone, we are fine.
--
MOS_MIGRATED_REVID=127703598
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127701838
|
|
|
|
|
|
|
| |
unlike System#currentTimeMillis().
--
MOS_MIGRATED_REVID=127697254
|
|
|
|
|
|
|
| |
Isn't strictly necessary, remove it before anyone decides they want to use it.
--
MOS_MIGRATED_REVID=127585617
|
|
|
|
|
|
|
|
|
| |
rather than a PRECOMPUTED value.
Having a stale TopLevelArtifactContext leads to invalidation of all the top level target nodes, causing time wasted due to a lot of cache hits for a null build.
--
MOS_MIGRATED_REVID=127585059
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Before this change:
Any given cc_library can only contribute one library with a given name to
targets which depend on it. If an input library has the same name as the
cc_library which it is an input to, the decision of which to use is based
on the link mode. e.g.,
cc_library(
name = "foo",
srcs = ["foo.c", "libfoo.so"],
)
will only contribute libfoo.a (a static library containing foo.o) in static mode,
while it will only contribute libfoo.so (the precompiled shared library) in dynamic
mode.
This change alters cc_library's behavior in this case:
* If libfoo.a would be empty (i.e., there are no linkable sources), then
this is allowed. The libfoo.so from srcs is simply passed through. (Previously,
the empty libfoo.a would be forwarded.)
* Otherwise, this is an error.
In the case where there are multiple libraries in the srcs with the same
library identifier (lib[name].[a|so|lo]), cc_library will still choose one
based on the link mode. This behavior has not changed.
Similarly, cc_library will still choose one of its own outputs based on the
link mode. That behavior has not changed either.
RELNOTES[INC]: It is now an error to include a precompiled library (.a, .lo, .so)
in a cc_library which would generate a library with the same name
(e.g., libfoo.so in cc_library foo) if that library also contains other linkable
sources.
--
MOS_MIGRATED_REVID=127569615
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127543896
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127538990
|
|
|
|
|
|
|
| |
--
Change-Id: I3255a14a60b7ae7749c49d5a885d92f4f19ec84f
Reviewed-on: https://bazel-review.git.corp.google.com/#/c/3980/
MOS_MIGRATED_REVID=127537367
|
|
|
|
|
|
|
|
|
| |
-XX:TieredStopAtLevel=1 instead of -client, as the latter is ignored on 64-bit JVMs since at least JDK 6.
The new flags have been shown to cut down CPU time by 50% due to reduced time spent in JIT and improving overall run time for short running actions, some times also cutting them in half.
--
MOS_MIGRATED_REVID=127536578
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127531480
|
|
|
|
|
|
|
| |
are already compiled into a proto runtime.
--
MOS_MIGRATED_REVID=127489419
|
|
|
|
|
|
|
| |
What we really are doing here is formatting.
--
MOS_MIGRATED_REVID=127481183
|
|
|
|
|
|
|
|
|
| |
Turns out that our existing test coverage was insufficient for catching this (I
misread the tests during development of unknown commit and thus incorrectly
thought we had test coverage for this).
--
MOS_MIGRATED_REVID=127461716
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127458867
|
|
|
|
|
|
|
|
|
|
| |
The new packaging rules need these in order to access resources included in
objc_library dependencies of an application.
RELNOTES:
--
MOS_MIGRATED_REVID=127445152
|
|
|
|
|
|
|
|
| |
select() as "nonconfigurable", polish up general configurable
attributes docs.
--
MOS_MIGRATED_REVID=127440164
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127435666
|
|
|
|
|
|
|
|
|
|
|
| |
related to tests.
This allows for better caching and also makes it easy to run tests in a different directory than the global execroot.
The paths are rewritten to absolute paths in test-setup.sh, because Test Encyclopedia says that we pass absolute paths to tests in these variables.
--
MOS_MIGRATED_REVID=127432675
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127425448
|
|
|
|
|
|
|
|
|
| |
and commit 6388680b44893c2b3ccddd73f6c4abb693a18cd4 now that the incompatibilities have been resolved.
RELNOTES[INC]: The host_platform and target_platform entries are not written to the master log anymore.
--
MOS_MIGRATED_REVID=127417109
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Broke Android targets with native code when cpu is explicitly set.
*** Original change description ***
RELNOTES: Improve Android split transition handling.
--
MOS_MIGRATED_REVID=127377943
|
|
|
|
|
|
|
| |
The classpaths get split during flag parsing, so joining them is unnecessary.
--
MOS_MIGRATED_REVID=127377219
|
|
|
|
|
|
|
|
|
|
| |
Saves some needless garbage generation. We expect TestSize#getTestSize(String)
to always pass in a lowercase value, which we promptly change to uppercase for
the enum lookup. Instead memoize the lowercase -> TestSize mapping to avoid
this call.
--
MOS_MIGRATED_REVID=127362499
|
|
|
|
|
|
|
|
|
|
|
| |
and use it to work around an analysis performance regression caused by header
compilation. SpawnAction expands the full argument list and compares the length
to --min_param_file_size when deciding whether to use a params file, and header
compilation actions often have very long JoinExecPathsArg fragemnts that are
expensive to expand.
--
MOS_MIGRATED_REVID=127354241
|
|
|
|
|
|
|
|
|
|
|
| |
ThinLTO isn't fully supported for host builds, in order to handle this add
support for enabling either a 'host' or 'target' feature in the configuration.
The thin_lto feature will require the new 'target' configuration.
Companion crosstool change: unknown commit.
--
MOS_MIGRATED_REVID=127350506
|
|
|
|
|
|
|
| |
This is equivalent to the deprecation of NestedSetBuilder#addAll(NestedSet).
--
MOS_MIGRATED_REVID=127336417
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=127334955
|