When the --rename-resources-package flag is used, only rename the
package name within the resources.arsc. Generate the R.java under
the package name defined by the manifest unless --custom-package is
used.
(This is a cherry pick.)
Bug: 147434671
Test: manual
Merged-In: Id9d10e16f32533da1b67ab72ac846791526e7a76
Change-Id: Id9d10e16f32533da1b67ab72ac846791526e7a76
There is currently no way to change the package name encoded in the
resources table. This change adds a flag that controls rewriting the
package name encoded into the resources table. This flag does not
rewrite the package name in the manifest xml.
(This is a cherry pick.)
Bug: 147434671
Test: manual
Merged-In: I435ba07e29df0522c3a44cc5f2c460004a02d71c
Change-Id: I435ba07e29df0522c3a44cc5f2c460004a02d71c
Bug: 154760495
Test: make with WITH_TIDY=1 and DEFAULT_GLOBAL_TIDY_CHECKS=clang-analyzer-core.uninitialized.Branch
Change-Id: I569749166177c25e63b32fb468dccd13797acc05
Preparation for removing implicit include paths for jni.h from soong.
Also remove unnecessary jni.h includes from PathParser.{h,cpp}.
Bug: 152482542
Test: lunch aosp_x86_64 && m checkbuild
Change-Id: Ib18afa3d2c93ed3ce69204286d5177991100fd7e
Exempt-From-Owner-Approval: build refactoring
* normally a copy constructor should call base copy constructor
Bug: 151109313
Test: WITH_TIDY=1 make
Change-Id: I9677889973178c26cd787a457333877619956b51
We should keep classes that are in the android:name or class xml
attributes.
Test: m -j aapt2_tests
Bug: 142601969
Change-Id: Ia67365bd702bae75d38b9572d68e9930e856e0f8
- Allow passing list of explicit fieldnames, instead of sorting header names.
- Allow passing a single zip file with all CSV files to merge.
Bug: 145132366
Test: diff -w out/soong/hiddenapi/hiddenapi-greylist.csv
Change-Id: I7c6db369069452293c15eec290c635fb3879275d
- Use csv.QUOTE_MINIMAL strategy to only quote fields with special characters.
- Open file as 'r', not 'rb'.
- Use 'unix' dialect for '\n' as line terminator, instead of \r\n.
Bug: 145132366
Test: diff -w out/soong/hiddenapi/hiddenapi-greylist.csv
Change-Id: I888f3219beee5b83e53f05589a90a05f93399fe3
As part of getting rid of StatsLogInternal and moving every atom to be
logged from a module, this CL addresses 2 issues:
1. Add WorkSource logging to modules.
2. Gate WorkSource logging via a CLI flag to prevent WorkSource methods
from being included in modules that are Mainline modules and can't have
@hide calls.
Bug: 147541828
Test: m
Test: atest android.cts.statsd.atom.UidAtomTests
Change-Id: Icce2f6ed7298be5e5ceebfb05d87e6fa7cd6bdbb
Merged-In: Icce2f6ed7298be5e5ceebfb05d87e6fa7cd6bdbb
API Review requested to remove unneeded allocations so pass null for
arguments that are empty when writing KeyValuePairs
Bug: 146383532
Test: m
Change-Id: Ic1788f65e8c7f837d6800d59c186519a82882bf7
Merged-In: Ic1788f65e8c7f837d6800d59c186519a82882bf7
(cherry picked from commit 996afa633c)
Don't use Build.VERSION.CODENAME to check for R.
Use Build.VERSION.SDK_INT > Q || (Build.VERSION.SDK_INT == Q &&
Build.VERSION.PREVIEW_SDK_INT > 0)
Bug: 146235828
Test: m DocumentsUIGoogle
Change-Id: I59e327d4e92d41a93b6bea145b8497fd4f85108c
Merged-In: I59e327d4e92d41a93b6bea145b8497fd4f85108c
(cherry picked from commit b6f32d9fe2)
* changes:
Add --supportQ flag to java StatsLog codegen
Support new socket schema in native codegen
Move statsd metadata from statslog to atoms_info
Fix StatsEvent memory usage for pulled events
Before this CL, classes in the unnamed package, such as
L$r8$backportedMethods$utility$Objects$2$checkIndex
(which looks like it is generated by R8 for Objects.checkIndex() calls)
were breaking the build because extract_package was incorrectly assuming
that there would always be a '/' in the identifier string.
Test: Build that previously broke is now working.
Change-Id: Ice78d6b31c4f38a3c9d529bc6156d625d19bcacf
Right now, Q schema generated code has a dependency on libutils.
Turning on R schema for clients will remove libutils as a required
dependency. Once libutils is removed from clients, turning the flag off
again will be complicated as all clients will need to re-include
libutils. To fix this problem, remove dependency on libutils in Q
schema.
Right now, we only use SystemClock.h from libutils in Q schema autogen.
Use timespec from C standard library instead.
Bug: 146808265
Test: m
Test: flashes successfully
Test: adb logcat "*:S statsd:*"
Test: bit statsd_test:*
Change-Id: I00efeab1be351a95083f3b6e2dbc9067849bee20
Merged-In: I00efeab1be351a95083f3b6e2dbc9067849bee20
Add support to generate Q schema code generation only for modules that
ship to Q.
Bug: 145606209
Test: m
Test: flashes successfully
Test: adb logcat "*:S statsd:*"
Change-Id: I978ac321415f1e8d76e4ec73445894965845aaea
Merged-In: I978ac321415f1e8d76e4ec73445894965845aaea
Clients don't need to know about statsd metadata. Extract out metadata
from statslog.h/cpp into atoms_info.h/cpp which is only used by statsd.
Generated atoms_info.h: https://paste.googleplex.com/6303016724463616
Generated atoms_info.cpp: https://paste.googleplex.com/5717940978581504
Test: m -j
Test: old metadata in statslog matches the metadata in
atoms_info
Test: Flashes successfully
Test: adb logcat "*:S statsd:*"
Change-Id: I56ef3cc4ea1fbd2cd0130d4e9576b242efb9f627
Merged-In: I56ef3cc4ea1fbd2cd0130d4e9576b242efb9f627
Add usePooledBuffer flag to the Builder which determines whether to reuse
the Buffer's byte array in StatsEvent or use a copy.
The build() function also calls release() on the Buffer
if a copy of the Buffer's byte array is passed to StatsEvent.
Also, for pushed events, release the StatsEvent object and consequently,
the Buffer in StatsLog.write(StatsEvent)
Fixes: 145026572
Fixes: 144126444
Test: bit FrameworksCoreTests:android.util.StatsEventTest
Change-Id: I1cdaf0027b69281cb7cb6f3c8ca923d03829b4dd
Merged-In: I1cdaf0027b69281cb7cb6f3c8ca923d03829b4dd
The change is guarded by STATS_SCHEMA_LEGACY flag in the autogenerator
Android.bp.
The flag can only be removed once rest of statsd switches over to the
new socket schema.
The new autogeneration uses StatsEvent for both Mainline module
clients and the default platform client. For Mainline modules, the
autogenerated code uses a runtime dessert version check to use new
socket schema on R+ platforms and the legacy schema on the older
platforms.
The legacy schema is encapsulated in a separate autogenerated inner
class called QLogger.
generated DocumentsStatsLog.java: https://paste.googleplex.com/4665805503463424
generated StatsLogInternal.java: https://paste.googleplex.com/5955095055302656
generated android_util_StatsLogInternal.cpp:
https://paste.googleplex.com/6737331711115264
Bug: 142811546
Test: m -j && m -j DocumentsUIGoogle
Test: Flashes successfully and events are logged to statsd
Change-Id: I4c804eaf4d5ae78001146c89ebe46dfb0a453853
Merged-In: I4c804eaf4d5ae78001146c89ebe46dfb0a453853
"aapt2 compile" marks styleables as public instead of preserving
information from <public/> or --visibility options. This behavior can
now be disabled via --preserve-visibility-of-styleables.
Bug: 146649511
Change-Id: Ifb8ab396573d1393df737a59625e79e9cf2494a7
Tested: aapt2_tests
LocaleDataTables.cpp is outdated due to ICU update over the years
Test: ./tools/localedata/extract_icu_data.py $ANDROID_BUILD_TOP
Test: m droid
Change-Id: I21918cc0f50c7347e514dc7b015607e19cffb377
The property is currently set by the AIDL compiler to indicate the original
position of the annotation in the AIDL file. Use this position in
preference to the annotation position within Java code if it's present.
Also rework the tests to use the google compile-testing library rather than
my crufty javac library. This makes testing failure conditions much nicer.
Bug: 145120552
Test: atest unsupportedappusage-processor-test
Test: m framework-annotation-proc
Change-Id: Icaac9b40672dce028095b578c19950b688506c0d
If the signature has @UnsupportedAppUsage or @SystemApi annotation as well, those take precedence and the API would either be whitelisted or greylisted accordingly. All other, "pure" @TestApi signatures would be blacklisted and only allowed access in instrumented processes.
Test: manual
Bug: 133832325
Change-Id: I4684929caed8be7c42c91fed33ddd2a3b67ae19b
It seems the compiler has started adding a top level MODULE which broke
assumptions made in here.
Also add some tests to verify this and other basic functionality, so that
future such breaks may be caught sooner.
Test: m framework-annotation-proc-index
Test: atest unsupportedappusage-processor-test
Bug: 144495446
Change-Id: I43efb25c1600af15809d48aa84498a6800f713b1
The results of CollectLocations() are only used if the --proguard-conditional-keep-rules option is set, but the function is always called when the --proguard option is set.
Since this function can be slow on particularly large resource sets, avoid calling it unless --proguard-conditional-keep-rules is set.
Change-Id: If259d41cdab16abad8ca62f30d8beb5ebf33575a
Bug: 144236322
Test: aapt2_tests
When the protobuf header (or data payload) size are a multiple of 4,
"entry_length" (called "aligned size" in the code) would have counted bytes that
are never actually written.
Bug: 139418052
Change-Id: Ia688a82a67f3807f7feb0be03670bf2827b1d6a1
(cherry picked from commit b99e50922b)
