Can I multiply options in matrix.include[] with the env option? - travis-ci

Given a fairly heterogenous matrix that looks like this:
matrix:
include:
- os: linux
compiler: gcc
env: PLATFORM=android ARCH=arm64-v8a
- os: linux
compiler: gcc
env: PLATFORM=linux ARCH=aarch64
- os: osx
compiler: clang
env: PLATFORM=darwin ARCH=x86_64 TEST=unit
- os: osx
compiler: clang
env: PLATFORM=ios ARCH=arm64
This would result in four builds. I'd like to multiply it by two with an additional environment variable TYPE=Debug/Release. What's the best way to achieve this effect? Consider that I've only shown four configurations but the real number of configurations is 15. I'm hoping I won't have to duplicate everything twice.
I've tried the following but it merely adds two more builds, it doesn't combine with the matrix:
env:
matrix:
- TYPE=Debug
- TYE=Release
Same with this:
env:
- TYPE=Debug
- TYE=Release

I am afraid you cannot.
There is no matrix expansion that you might expect on the top level does not happen in matrix.include.
The env key forms just one axis of the build matrix, and it is not possible to use different env values to construct a build matrix.
The only way to achieve what you need is adding four more jobs:
matrix:
include:
- os: linux
compiler: gcc
env: PLATFORM=android ARCH=arm64-v8a TYPE=Debug
- os: linux
compiler: gcc
env: PLATFORM=linux ARCH=aarch64 TYPE=Debug
- os: osx
compiler: clang
env: PLATFORM=darwin ARCH=x86_64 TEST=unit TYPE=Debug
- os: osx
compiler: clang
env: PLATFORM=ios ARCH=arm64 TYPE=Debug
- os: linux
compiler: gcc
env: PLATFORM=android ARCH=arm64-v8a TYPE=Release
- os: linux
compiler: gcc
env: PLATFORM=linux ARCH=aarch64 TYPE=Release
- os: osx
compiler: clang
env: PLATFORM=darwin ARCH=x86_64 TEST=unit TYPE=Release
- os: osx
compiler: clang
env: PLATFORM=ios ARCH=arm64 TYPE=Release
(Incidentally, where are you getting the number 15?)

I usually use a small python script with a (jinja) template included to generate my travis config.
#!/usr/bin/env python3
tpl = '''
script:
- some_stuff.bash
matrix:
include:
{%- for letter, number in vars %}
- env: TESTS={{number}} CONFIG={{letter}}
{%- endfor %}
'''
import os
import jinja2
from itertools import product
tpl = jinja2.Template(tpl)
ymlfn = os.path.join(os.path.dirname(__file__), '.travis.yml')
with open(ymlfn, 'wt') as yml:
yml.write(tpl.render(vars=product(['a', 'b'], range(5))))
extend example

Related

The clang compiler does not support a feature it reportedly supports

I'm trying to tailor compiler flags of clang-11 to the apple-m1 CPU, which clang-11 doesn't know about yet.
The output of /usr/bin/clang -E - -mcpu=apple-m1 -### on macOS outputs a command with flags like these in it: "-target-feature" "+zcz"
From that you can infer the following features of the CPU:
armv8.5a+fp-armv8+neon+crc+crypto+dotprod+fp16fml+ras+lse+rdm+rcpc+zcm+zcz+fullfp16+sm4+sha3+sha2+aes
However, out of these, +fp-armv8+neon+zcm+zcz+fullfp16 are not recognised to be valid by any clang compiler:
$ cc -march=armv8.5a+zcz test.c
clang-11: error: the clang compiler does not support '-march=armv8.5a+zcz'
How can I tell clang to optimise for those target flags?

Qbs does not pick Clang in the tool chain configuration

I have installed LLVM 5.0 on Windows 10 x64. When I call clang++ --version from cmd the message below is displayed.
clang version 5.0.0 (tags/RELEASE_500/final)
Target: x86_64-pc-windows-msvc
Thread model: posix
InstalledDir: C:\LLVM\bin
But when I call qbs setup-toolchains --detect in the resulting list neither clang nor clang++ is in. If I call qbs config --list profiles both clang and clang++ are not in the list. Am I missing something?
On Windows, the --detect option only looks for MSVC and mingw installations. Use the explicit mode instead:
qbs setup-toolchains C:\LLVM\bin\clang.exe clang

glibcxx_3.4.20 not found

I'm using Ubuntu 14.04, after installing LLVM & Clang, when i tap this in the terminal it reports the error:
wishfay#wishfay-virtual-machine:~$ clang -v
clang: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version `GLIBCXX_3.4.20' not found (required by clang)
And i check my libstdc++.so.6:
wishfay#wishfay-virtual-machine:~$ locate libstdc++.so.6
/usr/lib/vmware-tools/lib32/libstdc++.so.6
/usr/lib/vmware-tools/lib32/libstdc++.so.6/libstdc++.so.6
/usr/lib/vmware-tools/lib64/libstdc++.so.6
/usr/lib/vmware-tools/lib64/libstdc++.so.6/libstdc++.so.6
/usr/lib/x86_64-linux-gnu/libstdc++.so.6
/usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.19
/usr/share/gdb/auto-load/usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.19-gdb.py
And my gcc version:
wishfay#wishfay-virtual-machine:~$ gcc --version
gcc (Ubuntu 4.8.4-2ubuntu1~14.04.3) 4.8.4
And libstdc++.so.6:
wishfay#wishfay-virtual-machine:~$ strings /usr/lib/x86_64-linux-gnu/libstdc++.so.6 | grep GLIBCXX
GLIBCXX_3.4
GLIBCXX_3.4.1
GLIBCXX_3.4.2
GLIBCXX_3.4.3
GLIBCXX_3.4.4
GLIBCXX_3.4.5
GLIBCXX_3.4.6
GLIBCXX_3.4.7
GLIBCXX_3.4.8
GLIBCXX_3.4.9
GLIBCXX_3.4.10
GLIBCXX_3.4.11
GLIBCXX_3.4.12
GLIBCXX_3.4.13
GLIBCXX_3.4.14
GLIBCXX_3.4.15
GLIBCXX_3.4.16
GLIBCXX_3.4.17
GLIBCXX_3.4.18
GLIBCXX_3.4.19
GLIBCXX_DEBUG_MESSAGE_LENGTH
I want to know how can i get the GLIBCXX_3.4.20 .
According to the ABI page, GLIBCXX_3.4.20 is part of gcc-4.9.
You need to upgrade your version of gcc, or get a clang binary that supports your OS, or you need to build clang from source.

Cross Compile OpenCV for Rpi2 with Java bindings

What I try to do is to cross compile OpenCV from a x86 host machine to an ARM target machine (Raspberry Pi 2) with Java bindings.
All I've achieved is to compile OpenCV with Java bindings for x86 platform, or even OpenCV with NO Java bindings for ARM platform. However I cannot compile OpenCV with Java bindings for ARM platform.
I've kind of followed thousands of guides to do this. This is from OpenCV's official site, and seems to be very simple: http://docs.opencv.org/2.4/doc/tutorials/introduction/crosscompilation/arm_crosscompile_with_cmake.html
My host machine is the following:
$ uname -a:
Linux ubuntu 4.2.0-16-generic #19-Ubuntu SMP Thu Oct 8 14:46:51 UTC 2015 i686 i686 i686 GNU/Linux
I've installed gcc and g++ cross compilation tools (gnueabi and gnueabihf):
$ sudo apt-get install gcc-arm-linux-gnueabi
$ sudo apt-get install g++-arm-linux-gnueabi
$ sudo apt-get install gcc-arm-linux-gnueabihf
$ sudo apt-get install g++-arm-linux-gnueabihf
$ which arm-linux-gnueabihf-gcc
/usr/bin/arm-linux-gnueabihf-gcc
$ which arm-linux-gnueabihf-g++
/usr/bin/arm-linux-gnueabihf-g++
Since I want to compile OpenCV with the Java bindings, I installed jdk and ant:
$ sudo apt-get install openjdk-7-jre
$ sudo apt-get install openjdk-7-jdk
$ sudo apt-get install ant
Then I add these lines to .bashrc file:
JAVA_HOME=/usr/lib/jvm/java-7-openjdk-i386
PATH=$JAVA_HOME/bin:$PATH
export PATH
Then I execute:
$ source $HOME/.bashrc
I've downloaded OpenCV's source code and moved to the platforms/linux folder as the official guide does:
$ cd ~/opencv/platforms/linux
$ mkdir -p build_hardfp
$ cd build_hardfp
Then in the "opencv/platforms/arm-gnueabi.toolchain.cmake" file, replaced these lines:
set(CMAKE_C_COMPILER arm-linux-gnueabi${FLOAT_ABI_SUFFIX}-gcc-${GCC_COMPILER_VERSION})
set(CMAKE_CXX_COMPILER arm-linux-gnueabi${FLOAT_ABI_SUFFIX}-g++-${GCC_COMPILER_VERSION})
by these:
set(CMAKE_C_COMPILER /usr/bin/arm-linux-gnueabihf-gcc)
set(CMAKE_CXX_COMPILER /usr/bin/arm-linux-gnueabihf-g++)
In order to use arm compiler instead of x86's.
Then I try to get cmake ready:
$ cmake -DBUILD_SHARED_LIBS=OFF -D BUILD_NEW_PYTHON_SUPPORT=NO -DCMAKE_TOOLCHAIN_FILE=../arm-gnueabi.toolchain.cmake ../../..
Cmake's output tells it will use arm cross compiler for ARM platform target, but it will not use Java bindings:
...
-- Platform:
-- Host: Linux 4.2.0-16-generic i686
-- Target: Linux 1 arm
-- CMake: 3.2.2
-- CMake generator: Unix Makefiles
-- CMake build tool: /usr/bin/make
-- Configuration: Release
...
C++ Compiler: /usr/bin/arm-linux-gnueabihf-g++ (ver 5.2.1)
...
-- OpenCV modules:
-- To be built: core flann imgproc highgui features2d calib3d ml video legacy objdetect photo gpu ocl nonfree contrib stitching superres ts videostab
-- Disabled: world
-- Disabled by dependency: -
-- Unavailable: androidcamera dynamicuda java python viz
So, I try to set cmake compile variables by myself without using cmake's toolchain file:
$ export CMAKE_C_COMPILER=/usr/bin/arm-linux-gnueabihf-gcc
$ export CMAKE_CXX_COMPILER=/usr/bin/arm-linux-gnueabihf-g++
$ cmake -DBUILD_SHARED_LIBS=OFF -D BUILD_NEW_PYTHON_SUPPORT=NO ../../..
Now cmake's output tells it will include Java support, but it won't use arm cross compiler:
...
-- Platform:
-- Host: Linux 4.2.0-16-generic i686
-- CMake: 3.2.2
-- CMake generator: Unix Makefiles
-- CMake build tool: /usr/bin/make
-- Configuration: Release
...
C++ Compiler: /usr/bin/c++ (ver 5.2.1)
...
-- OpenCV modules:
-- To be built: core flann imgproc highgui features2d calib3d ml video legacy objdetect photo gpu ocl nonfree contrib java stitching superres ts videostab
-- Disabled: world
-- Disabled by dependency: -
-- Unavailable: androidcamera dynamicuda python viz
Of course, if I execute make command with this latest cmake configuration, this is the ".so" file I get:
$ readelf -h lib/libopencv_java249.so | grep Machine
Machine: Intel 80386
where it should tell:
Machine: ARM
So, once again: I can compile OpenCV with Java bindings for x86 platform, or either OpenCV with NO Java bindings for ARM platform, but not both.
How should I do this?
Thank you!
UPDATE 1:
#Notlikethat I forgot to tell I had already tried that (i.e. use ARM jdk instead of x86). I did not mention it because I though I should be using x86.
However, I have tried it again:
I've downloaded ARM's jdk, set JAVA_HOME and PATH variables properly to point this new jdk and tried cmake command.
The result is the same, it lets me compile for ARM without Java bingings, or for x86 with Java bindings.
UPDATE 2:
I've added the following variables to the "arm-gnueabi.toolchain.cmake" file:
set(JAVA_HOME /usr/lib/jvm/jdk1.7.0_60_ARM)
set(JAVA_AWT_LIBRARY $JAVA_HOME/include/jawt.h)
set(JAVA_JVM_LIBRARY $JAVA_HOME/jre/lib/arm/jvm.cfg)
set(JAVA_INCLUDE_PATH $JAVA_HOME/include/jni.h)
set(JAVA_INCLUDE_PATH2 $JAVA_HOME/include/linux/jni_md.h)
set(JAVA_AWT_INCLUDE_PATH $JAVA_HOME/include/jawt.h)
Now if I execute:
$ cmake -DBUILD_SHARED_LIBS=OFF -D BUILD_NEW_PYTHON_SUPPORT=NO -DCMAKE_TOOLCHAIN_FILE=../arm-gnueabi.toolchain.cmake ../../..
the output shows that java module is still unavailable, but at least, one of its key dependencies is ok (JNI):
...
-- Java:
-- ant: NO
-- JNI: $JAVA_HOME/include/jni.h $JAVA_HOME/include/linux/jni_md.h $JAVA_HOME/include/jawt.h
-- Java tests: NO
...
I'm pretty sure the problem here is the fact that ant is not found, which I can't understand.
Ant is installed:
$ echo $PATH:/usr/lib/jvm/jdk1.7.0_60_ARM/bin:/opt/apache/ant/apache-ant-1.9.6/bin:...
I've retried by adding the following variables to the "arm-gnueabi.toolchain.cmake" file, without success:
set(ANT_HOME /opt/apache/ant/apache-ant-1.9.6)
set(JAVA_ANT $ANT_HOME/bin/ant)

Pkg-config 0.28 did not build on OSX 10.10.2

Recently I clean installed Yosemite 10.10.2
I also installed gcc using
brew install gcc
Now I'm trying to build and install python and openmpi from source because previously I've had problems with gfortran and clang.
So I did
brew install openmpi --build-from-source -verbose
brew install python --build-from-source -verbose
with variables and aliases:
export CC=gcc-4.9
export CXX=g++-4.9
export CPP=g++-4.9
export LD=gcc-4.9
export FC=gfortran-4.9
export HOMEBREW_CC=gcc-4.9
export HOMEBREW_LD=gcc-4.9
export HOMEBREW_CXX=g++-4.9
export HOMEBREW_CPP=g++-4.9
export HOMEBREW_FC=gfortran
alias cc=gcc-4.9
alias gcc=gcc-4.9
alias c++=c++-4.9
alias g++=g++-4.9
Still getting the following error:
In file included from /usr/include/dispatch/dispatch.h:51:0,
from /System/Library/Frameworks/CoreFoundation.framework/Headers/CFStream.h:15,
from /System/Library/Frameworks/CoreFoundation.framework/Headers/CFPropertyList.h:13,
from /System/Library/Frameworks/CoreFoundation.framework/Headers/CoreFoundation.h:55,
from /System/Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:19,
from gunicollate.c:30:
/usr/include/dispatch/object.h:143:15: error: expected identifier or '(' before '^' token
typedef void (^dispatch_block_t)(void);
/usr/include/dispatch/object.h:362:3: error: unknown type name 'dispatch_block_t'
dispatch_block_t notification_block);
==> Formula
Path: /usr/local/Library/Formula/pkg-config.rb
==> Configuration
HOMEBREW_VERSION: 0.9.5
ORIGIN: https://github.com/Homebrew/homebrew
HEAD: 9bbd6637009256d70b57691c3ad1979ca2bb5916
Last commit: 9 hours ago
HOMEBREW_PREFIX: /usr/local
HOMEBREW_CELLAR: /usr/local/Cellar
CPU: dual-core 64-bit penryn
OS X: 10.10.2-x86_64
Xcode: 6.2
CLT: 6.2.0.0.1.1424975374
Clang: 6.0 build 600
X11: N/A
System Ruby: 2.0.0-p481
Perl: /usr/bin/perl
Python: /usr/bin/python
Ruby: /usr/bin/ruby
Java: 1.6.0_65
==> ENV
HOMEBREW_CC: gcc-4.9
HOMEBREW_CXX: g++-4.9
MAKEFLAGS: -j2
CMAKE_PREFIX_PATH: /usr/local
CMAKE_INCLUDE_PATH: /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk/usr/include/libxml2:/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk/usr/include/apache2:/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk/System/Library/Frameworks/OpenGL.framework/Versions/Current/Headers
CMAKE_LIBRARY_PATH: /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk/System/Library/Frameworks/OpenGL.framework/Versions/Current/Libraries
CMAKE_FRAMEWORK_PATH: /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk/System/Library/Frameworks
PKG_CONFIG_LIBDIR: /usr/lib/pkgconfig:/usr/local/Library/ENV/pkgconfig/10.10
HOMEBREW_SDKROOT: /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk
PATH: /usr/local/Library/ENV/4.3:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/opt/gcc/bin
Error: pkg-config 0.28 did not build
My understanding is that there is a conflict between clang and gcc-4.9 configuration.
I could temporally solve it doing:
brew install pkg-config
So it doesn't build if from source and uses clang configuration.
I see that Xcode is still in the Include and Library path, and I think this prevents linking with gcc-4.9.
How can I install pkg-config using gcc-4.9 only?
Problem solved after following this instructions changing:
typedef void (^dispatch_block_t)(void); // this is for clang
To
typedef void* dispatch_block_t; //this is for gcc-4.9

Resources