KuduPredicate::Clone on IN list predicate causes segfault
Jordan Birdsell uncovered this issue where a cloned IN list predicate
causes a segfault when applied to a scanner. During the clone, the
values list was being 'preallocated' using the vector(int) constructor,
which adds a bunch of default c
[docs] Cleanup beta mentions, links
Change-Id: I56fa65be9d027f5b40628aa5db5f9e4a57c1fbc9
Reviewed-on: http://gerrit.cloudera.org:8080/4565
Reviewed-by: David Ribeiro Alves
Tested-by: Kudu Jenkins
Project: http://git-wip-us.apache.org/repos/asf/kudu/repo
Commit: http://git-wip-us.apache.org/repo
build-and-test.sh: update gcovr location
Coverage builds were failing because of this.
Change-Id: I8874d298b6176b4deb45749847032e5c2be55955
Reviewed-on: http://gerrit.cloudera.org:8080/4567
Reviewed-by: David Ribeiro Alves
Tested-by: Adar Dembo
Project: http://git-wip-us.apache.org/repos/asf/
KUDU-1657: read-only FsManager::Open on active tablet can crash
This fixes a race condition when opening a read-only FsManager on a data
directory with lbm containers that are actively being appended to. See
the code comment for more details. A best-effort repro test case is
included; it typically
[tools] added insert-generated-rows into kudu tools
The insert-generated-rows tool has been merged into the 'kudu'
umbrella toolset.
This addresses KUDU-1628.
Besides, added ability to run multiple inserter threads and
specify additional parameters on batching behavior of the generated
write ope
Repository: kudu
Updated Branches:
refs/heads/master 33e98ce49 -> 70211f84c
[docs] Release note for OperationResponse.getWriteTimestamp
See https://gerrit.cloudera.org/#/c/4487/
Change-Id: I6f81b1c70bc8f8be7f5529762ac9a9622dec00f0
Reviewed-on: http://gerrit.cloudera.org:8080/4560
Reviewed-by:
c++ client: stop requiring the old gcc ABI
With the upgrade to clang 3.9 and the transition to libc++ for TSAN, we can
now safely remove the old gcc ABI guard rail without breaking TSAN builds.
The impact on backwards compatibility is immaterial. At least one Kudu
vendor has shipped a client pack
thirdparty: build C dependencies with TSAN instrumentation too
Our C library dependencies are pretty small so this isn't a huge hit, and
it makes the sanitizer-based dependency stack much more consistent. Now,
'common' is only for tools and header-only dependencies, while
uninstrumented/tsan/... i
Repository: kudu
Updated Branches:
refs/heads/master cb92799ba -> 33e98ce49
thirdparty: reorganize tree
This patch changes the organization of the thirdparty tree. The new layout
looks like this:
- installed: _all_ installed dependencies, with 'common',
'uninstrumented', and 'tsan
thirdparty: fix up libtool scripts if needed
Older versions of libtool (such as the version found on el6) do not pass
-stdlib=libc++ found in CXXFLAGS down to the libtool link command line, and
as a result, the shared object is linked against the system libstdc++. The
net effect is that some third
thirdparty: split into dependency groups
The monolithic thirdparty build is now quite a bit larger than it used to be
on account of the extra LLVM build. Let's see if we can't speed it up. The
idea is simple: carve it up into disjoint sections so that individual
sections can be rebuilt as needed.
thirdparty: stifle unused argument warnings when building with clang
Dan tested my various thirdparty patches on macOS and reported that clang
emits thousands of unused argument warnings when building llvm, gflags, and
gtest. The first is due to the use of EXTRA_LDFLAGS in the llvm build; the
othe
thirdparty: use libc++ instead libstdc++ for TSAN builds
This all began because I wanted two things:
1. To use the new gcc 5 ABI on platforms that default to it (such as Ubuntu
Xenial). Other applications compiled on these platforms will use the new
ABI, and the fact that the Kudu client for
13 matches
Mail list logo
