| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
|
| |
Moved the last remaining logic to its corresponding support class
(XcodeSupport).
--
MOS_MIGRATED_REVID=93462899
|
|
|
|
|
|
|
| |
an attribute assignment.
--
MOS_MIGRATED_REVID=93460093
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently the ObjcBundleLibrary#create() function creates a
BundleSupport object with the default "iphone" target family.
An ObjcBundleLibrary is a descendant of a Bundle, which has
a "families" attribute, so the information is available.
Currently, ObjcBundleLibrary will unconditionally use "iphone"
as the target device family.
This commit updates ObjcBundleLibrary to use the "families"
attribute. It does this by making BundleSupport
control the "families" attribute, and then provide access
to it for the other classes that use it, like
ObjcBundleLibrary. The other classes are changed to
get their target device family information from their
BundleSupport objects.
--
Change-Id: I4a5be2849315dabfe7f2975c006a2092a56027af
Reviewed-on: https://bazel-review.googlesource.com/#/c/1260/3
MOS_MIGRATED_REVID=93456579
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93454041
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
coarsely checking for duplicates *anywhere*, e.g.:
select({':a': ['a.cc'], ':b': ['a.cc']}) + select({':a': ['b.cc'], ':b': ['b.cc']})
would fail. But this case is okay because these duplicates are in mutually
exclusive select paths (so they could never appear together anyway).
The new checking logic is:
- Duplicates can appear in different paths of the same select.
- Duplicates can *not* appear within the same path of a select.
- Duplicates can *not* appear across multiple selects (no matter what path -
this is still stricter than we need to be, but there's no strong case
for refining this case now).
--
MOS_MIGRATED_REVID=93447979
|
|
|
|
|
|
|
| |
tools/objc/BUILD.
--
MOS_MIGRATED_REVID=93447039
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When an ios_application is built, Bazel now collects the .breakpad files for
each architecture and tweaks the first line of each file to reflect the name of
the app, instead of the name of the binary:
MODULE mac arm64 5A70922132B738E194DDBC456F90F43F0 PrenotCalculatorBinary_bin
becomes
MODULE mac arm64 5A70922132B738E194DDBC456F90F43F0 PrenotCalculator
Bazel also correctly renames the files to end with their architecture, so that
the crash server can use them:
armv7/.../PrenotCalculatorBinary.breakpad
becomes
PrenotCalculator_armv7.breakpad
--
MOS_MIGRATED_REVID=93419816
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93417885
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93416921
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93415666
|
|
|
|
|
|
|
|
|
| |
This was found by building 'client' target with -Wall.
--
Change-Id: If14355813b83b9e29b36411eaf597de8d57bda6e
Reviewed-on: https://bazel-review.googlesource.com/#/c/1290
MOS_MIGRATED_REVID=93415497
|
|
|
|
|
|
|
|
|
|
|
| |
TestStrategy#getLocalRunfilesDirectory(...) was using manifest names
to infer if the runfiles symlinks were created by comparing the name
of manifests. This is a bit confusing and adds unnecessary reliance
on manifests which we'd like to reduce. Instead get whether or not
we created the symlinks from the boolean that determines it.
--
MOS_MIGRATED_REVID=93414150
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93413161
|
|
|
|
|
|
|
| |
different attributes, to which we have attached aspect.
--
MOS_MIGRATED_REVID=93412457
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93401974
|
|
|
|
|
|
|
| |
Note that this test still fails on my machine.
--
MOS_MIGRATED_REVID=93401330
|
|
|
|
|
|
|
|
|
| |
If otherTop did not enqueue its dep for evaluation before errorKey threw, its dep would never be enqueued and therefore never signal. This used to be ok because we didn't shut the threadpool down as aggressively.
The fix is just to delay the actual throwing of errorKey until otherTop is signaled.
--
MOS_MIGRATED_REVID=93349168
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93332951
|
|
|
|
|
|
|
| |
XCode project files. Bazel-generated XCode projects contain all necessary header search paths to resolve header imports. Header map is not needed.
--
MOS_MIGRATED_REVID=93329329
|
|
|
|
|
|
|
| |
upcoming release.
--
MOS_MIGRATED_REVID=93328637
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
every possible value an attribute can take) for attributes with
multiple selects:
Given
attr = select({':a': 'w', ':b': 'x'}) + select({':a': 'y', ':b': 'z'}
the naive approach is to combine every possible value of the first select
with every possible value of the second select (producing 4 possible values
from 2^2 visitations). But since these selects have the same exact
conditions, only two values are actually possible ("wy", "xz") from 2
visitations.
This change efficiently considers that case. More generally, given n
concatenated selects with the same conditions, it brings evaluation time
down from O(2^n) to O(n) (assuming two conditions per select). It also
works for partial matches: given a concatenation of 6 selects where
1, 3, and 5 have the same conditions and 2, 4, and 6 have the same conditions,
evaluation time goes from O(2^6) to O(2^2).
--
MOS_MIGRATED_REVID=93325787
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93314155
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93308663
|
|
|
|
|
|
|
| |
Added skyframe and foundations test_suite
--
MOS_MIGRATED_REVID=93308651
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93306731
|
|
|
|
|
|
|
|
|
|
|
| |
of hard-coding it and move the incremental deployment directory from /sdcard/incrementaldeployment to /data/local/tmp/incrementaldeployment.
The first is necessary because if the app is not run under the default user profile (e.g. Android For Work), the default location is not writable, thus, sadness, the second is necessary because /sdcard is not accessible from the non-default profile.
This also makes it possible to eventually install .so files there, because the +x bit can be set under /data/local/tmp (unlike under /sdcard).
--
MOS_MIGRATED_REVID=93287264
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93170106
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The -extdirs flag sets a search path to the directory containing the JDK's
extension classes, and should always be set explicitly when cross-compiling. If
it is unset, the ext directory of the host JDK will be used instead.
javac requires that -extdirs be a list of directories. The blaze --javac_extdir
flag takes the label of a filegroup with the 'path' attribute set, and passes
the directory path through to JavaBuilder.
RELNOTES: N/A
--
MOS_MIGRATED_REVID=93147656
|
|
|
|
|
|
|
| |
--
Change-Id: I523f31cf2179521847f644c426235bdf76e9d497
Reviewed-on: https://bazel-review.googlesource.com/#/c/1262/
MOS_MIGRATED_REVID=93146490
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The commit this reverts caused generated Xcode projects to fail to
find their files, as apparently the project.pbxproj file format does
not expand $() variables in the mainGroup element. This commit simply
restores the original xcodeproj generation logic, so files are found
by Xcode.
--
Change-Id: Id3cc57498384cc212ef41c30e9c5a4e5d3065e42
Reviewed-on: https://bazel-review.googlesource.com/#/c/1120
MOS_MIGRATED_REVID=93140923
|
|
|
|
|
|
|
|
|
| |
'x{key}x'.format(key = 2) == 'x2x'
It is a very simplified version of the Python format.
--
MOS_MIGRATED_REVID=93130656
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93129861
|
|
|
|
|
|
|
|
|
|
|
| |
Poorly performing features of the API have been deprecated in favor of better alternatives:
- use addZip(File) over addZip(InputStream) or addZip(String, InputStream)
- use addFile(ZipFileEntry) over addFile(String, Date, InputStream, DirectoryEntryInfo)
New zip package for high performance ZIP file manipulation. Can directly work with compressed ZIP entry data and has support for Zip64 (forces Zip32 by default).
--
MOS_MIGRATED_REVID=93128639
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Tested on Linux with the following command line:
$ bazel help info-keys | sort | uniq -c
And compared the output before and after.
Fixes #175
--
Change-Id: Ia879796abf6f5b6b5742bfc9574d64fe53a650a3
MOS_MIGRATED_REVID=93127869
|
|
|
|
|
|
|
| |
Fixes #175.
--
MOS_MIGRATED_REVID=93127118
|
|
|
|
|
|
|
|
|
|
|
|
| |
The one we need is already in option_processor.cc.
Tested on Linux with the following command line:
$ ./bootstrap_test.sh all
--
Change-Id: I5cdd781d0c2e20242773db1ccb7e9eccd2980a78
MOS_MIGRATED_REVID=93122461
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I forgot to look at the header file when doing
https://github.com/google/bazel/commit/a0592a13c8801f395a17641579f16a3c3bc4587e.
Now I ran the following command line to make sure I didn't miss any:
$ git grep seperated
--
Change-Id: Id834f87667135e91bd0c3024e0a99e52954a7038
MOS_MIGRATED_REVID=93120398
|
|
|
|
|
|
|
| |
Python.
--
MOS_MIGRATED_REVID=93120393
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93110537
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93066384
|
|
|
|
|
|
|
|
|
| |
We wind up combining them anyway in the manifest file and it seems that
this is consistent with how we read them back. Return them all at once
to avoid tedious duplication in handling of them.
--
MOS_MIGRATED_REVID=93058482
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93051778
|
|
|
|
|
|
|
| |
/bin/bash clobbers euid with uid, which is undesirable when process-wrapper is a setuid binary being used for isolation.
--
MOS_MIGRATED_REVID=93051178
|
|
|
|
|
|
|
| |
Better fix for #174.
--
MOS_MIGRATED_REVID=93045666
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The stat struct's size field is off_t, which varies in size between 32- and
64-bit systems. The size field in FileStatus.java is a long (which in Java
is always 64 bits) so the size is getting initialized incorrectly.
I can't think of a good way to add this to our integration tests, but I've
tested in manually on a VM.
Fixes #174.
--
MOS_MIGRATED_REVID=93038537
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93026026
|
|
|
|
|
| |
--
MOS_MIGRATED_REVID=93025992
|
|
|
|
|
|
|
|
| |
s/seperated/separated.
--
Change-Id: I303e4ff20b241c2c58fe00ad0a5e340b5bff54d9
MOS_MIGRATED_REVID=93025985
|
|
|
|
|
|
|
| |
Fixes #164.
--
MOS_MIGRATED_REVID=93025634
|
|
|
|
|
|
|
| |
defaults package.
--
MOS_MIGRATED_REVID=93023029
|