Each library in ardupilot is a folder with the source in. That avoids build
failures when there are files in libraries/. That fixes#4099 ("waf doesn't
seem to like TAGS files") and #4093 ("sim_vehicle.py /w waf /w eclipse project
broken on Cygwin").
That problem can be reproduced by just creating a file in libraries/ and trying
to build with waf.
There should be a better way to confirm a path is an ardupilot library. That
can be done later.
This is a hackish way to allow waf running on windows. In some
combinations of the tools the python interpreter seems to be crashing on
windows:
Found 180 MAVLink message types in 2 XML files
Generating C implementation in directory /tmp/ArduPlane.build/libraries/GCS_MAVLink/include/mavlink/v1.0/ardupilotmega
Generating C implementation in directory /tmp/ArduPlane.build/libraries/GCS_MAVLink/include/mavlink/v1.0/common
Copying fixed headers
last line in mavgen.py
Aborted (core dumped)
michael@WIN-3NBOUTHN4TN /cygdrive/c/Users/michael/Desktop/DIYDrones/ardupilot/ArduPlane
$ echo $?
134
Here we check the return code to be greater than 128 since that means the
interpreter received a signal. In this case we clear the return code.
Fix warnings like this:
<command-line>:0:21: warning: "APM_BUILD_DataFlash_test" is not defined [-Wundef]
/home/lucas/p/dronecode/ardupilot/libraries/AP_Vehicle/AP_Vehicle_Type.h:36:41: note: in expansion of macro ‘APM_BUILD_DIRECTORY’
#define APM_BUILD_TYPE(type) ((type) == APM_BUILD_DIRECTORY)
^
These happen because we are trying to set APM_BUILD_DIRECTORY to undefined
values. We should rather default to the APM_BUILD_DIRECTORY ==
APM_BUILD_UNKNOWN
check for the presence of libiio to enable the compilation of
the bebop rangefinder that needs it.
If the build remains static, there needs to be a rootfs that contains
the libiio.a file because it is not included in the debian package.
A solution can be to compile libiio and copy libiio.a in /usr/lib[/arch]
Otherwise it fails to link:
[335/335] Linking build/sitl/bin/arducopter-quad.exe
ArduCopter/libArduCopter_libs.a(SIM_Aircraft.cpp.4.o):SIM_Aircraft.cpp:(.text$_ZN4SITL8AircraftC2EPKcS2_+0x230): undefined reference to `_imp__timeGetTime@0'
ArduCopter/libArduCopter_libs.a(SIM_Aircraft.cpp.4.o):SIM_Aircraft.cpp:(.text$_ZN4SITL8Aircraft16setup_frame_timeEff+0x9f): undefined reference to `_imp__timeGetTime@0'
ArduCopter/libArduCopter_libs.a(SIM_Aircraft.cpp.4.o):SIM_Aircraft.cpp:(.text$_ZN4SITL8Aircraft16setup_frame_timeEff+0x125): undefined reference to `_imp__timeGetTime@0'
ArduCopter/libArduCopter_libs.a(SIM_Aircraft.cpp.4.o):SIM_Aircraft.cpp:(.text$_ZN4SITL8Aircraft15sync_frame_timeEv+0x21): undefined reference to `_imp__timeGetTime@0'
ArduCopter/libArduCopter_libs.a(SIM_Aircraft.cpp.4.o):SIM_Aircraft.cpp:(.text$_ZN4SITL8Aircraft15sync_frame_timeEv+0x152): undefined reference to `_imp__timeGetTime@0'
ArduCopter/libArduCopter_libs.a(SIM_Aircraft.cpp.4.o):SIM_Aircraft.cpp:(.text$_ZNK4SITL8Aircraft16get_wall_time_usEv+0xe): more undefined references to `_imp__timeGetTime@0' follow
collect2: error: ld returned 1 exit status
Before:
Checking for code snippet : yes
Checking for code snippet : yes
Checking for code snippet : yes
Checking for code snippet : yes
Checking for code snippet : no
Checking for code snippet : no
After:
Checking for HAVE_CMATH_ISFINITE : yes
Checking for HAVE_CMATH_ISINF : yes
Checking for HAVE_CMATH_ISNAN : yes
Checking for NEED_CMATH_ISFINITE_STD_NAMESPACE : yes
Checking for NEED_CMATH_ISINF_STD_NAMESPACE : no
Checking for NEED_CMATH_ISNAN_STD_NAMESPACE : no
The usual name for this header is config.h, but that's already used by
vehicles. Using uppercase could give the impression this is a
file to be modified, but it's not. Use lowercase instead.
Files that are not really part of the ROMFS in the folder might cause problems.
One problem that motivated this patch was caused because the make-based build
system copies the bootloader to the ROMFS in the source tree (mk/PX4/ROMFS)
instead of the build tree. That potentially could cause race condition between
the tasks created by 'px4_romfs_static_files' and 'px4_romfs_bootloader'.
Also, now we have only one task generator for static files.
That way we don't force other programs to be built on a directory of their
program group name. The directory name defaults to the program group.
We are separating those two concepts because of the upcoming support for
multiple groups for a program.
The find_realexec_path function was used for finding the toolchain path mostly
because of two reasons:
1) We couldn't really use CXX or CC variables because the user could set those
from the OS's environment and Waf wouldn't look for the executable file in
that case.
2) Our CI configuration sets up symlinks for ccache and find_realexec_path
works around that issue.
The bad side about using find_realexec_path() is that, besides working aroung
symlinks, it does the same thing that is done by Waf. This patch removes the
dependency for such a function by addressing each of the reasons above stated:
1) We create a local copy of os.environ and, if there's a variable with the
same name we are using, we remove it from the local copy.
2) As done before, we are looking for the cross ar program instead of gcc
program, since that is not used for ccache symlinks.
This is a better approach than checking command line options
--check-cxx-compiler and --check-c-compiler. Those values expect a list of
compilers to try instead of the compiler to use.
The benefits of this approach are:
- Allowing correct use of options --check-cxx-compiler and --check-c-compiler.
- Allowing user to pass CXX and CC environment variables, which is a common
way of selecting the compiler.
- Configuration is done *and committed* only for the specific compiler.
This is a better approach than checking command line options
--check-cxx-compiler and --check-c-compiler. Those values expect a list of
compilers to try instead of the compiler to use.
The benefits of this approach are:
- Allowing correct use of options --check-cxx-compiler and --check-c-compiler.
- Allowing user to pass CXX and CC environment variables, which is a common
way of selecting the compiler.
- Configuration is done *and committed* only for the specific compiler.
It makes more sense the toolchain Waf tool to be responsible of loading the
compilers. Furthermore, that allows toolchain tool to have control on doing
configuration before and after loading compiler tools.
- Use early return and reduce one indentation level.
- Set AR for both GNU compilers and clang just once and reduce redundancy.
- Reduce indentation level for clang-specific setup. There's no need to nest it
inside check if compilers are GNU or clang.
Copy the program library to the place where PX4Firmware CMake build will look
for and trigger the firmware build. There isn't really a real output defined
for px4_copy_lib because Waf would complain about multiple tasks having same
output when building multiple programs.
After the firmware build, copy it to the correct place (from program group and
name perspective) and add git hashes.
Since the place where the library is placed is shared by different target
programs, we need to synchronize the firmware build, that's why the use of
_firmware_semaphorish_tasks. That variable is set as a list because of the
upcoming upload task.
Recursively collect objects from dependency libraries and create a single
library. That way we just need to pass down one single library to PX4Firmware
build system.
In the upcoming build for PX4 boards, we will pass down the program a single
static library to PX4 Firmware's cmake build system. This patch is partially
providing a way to do that: the configuration for PX4 will define the
AP_PROGRAM_AS_STLIB environment variable.
The function ap_program() was the only one that was using it, so let's just
inline it. Besides, the name was misleading, since the (only) feature added was
for programs instead of general task generators.
Two things are fixed with this patch:
1. We sort dictionaries' keys if they aren't OrderedDict instances. Since
dict objects don't guarantee order, environment variables may have contents
in wrong order, causing unnecessary rebuilds.
2. We only use prepend_value() if there's already a value set for the key
and that value is list. Before this change, boards couldn't set
non-iterable values.
That will make platform specific naming be ignored. We use a string instead, to
let Waf tweak the target name correctly for us. The '#' prefix is to tell Waf
that the path is relative to bld.bldnode (instead of bld.path, which is the
default).
We need to remove CMakeCache.txt, otherwise cached variables would remain the
old value when they are removed from cmake_vars parameter.
We use `os.remove()` instead of `Node.delete()` because the latter removes the
node instance from its parent's children list. That makes the node be ignored
when storing persistent information after the build, thus the node signature
wouldn't be saved with that approach, which would make waf always think that
the task should be executed.
The default implementation takes into account the task's output and input
nodes. That isn't very well applicable for cmake tasks, so we define proper
uid() methods.
That allows options being declared where they're really used. Additionally, we
load ardupilotwaf after the other tools so that we can create our groups after
all non-ardupilot option groups are created. That makes our groups appear as
the last ones in the help message, which makes it easier to locate them.
The gtest header uses lots of undefined macros, showing lots of warnings
if we enable -Wundef. Ideally we could use a #pragma to ignore the
warning only from the correct header, but this currently doesn't work
with g++ - see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53431
So for now we disable the warning completely when compiling gtest or any
test that uses its header.
Thanks Gustavo Sousa
Copy the missing warnings from AP_Common.h and reorder the warnings to
be more clear on intent. This will later let us remove the warnings from
the header.
By running waf with option -v, waf was complaining about the update_submodule
tasks having the same identifier. That happened because the default uid()
method depended on the class name, inputs and outputs. Since this type of task
doesn't have input nor output file, we need to override uid() so that it uses
the submodule path for the task identifier.
With this change, there's no need to verify if the submodule is initialized,
because the submodules in GIT_SUBMODULES are automatically initialized and
updated if necessary before the build tasks are performed.
One downside of this change is that Google Benchmark configuration is now done
only during build. However that is minor, since now there are easy ways to
separately build different targets and program groups, so that a fail in
benchmark build doesn't really affect the other targets.
That is a workaround so that one can use `waf bin tests`, for example. In our
context, the only restriction to the task generators creation is values defined
during configuration. Thus, ideally, the recursion and task generators creation
should be done only once for multiple build calls. We should do that in the
future.
Gustavo Jose de Sousa
Francisco Ferreira
Andrew Tridgell
Lucas De Marchi
Julien Beraud
Ricardo de Almeida Gonzaga
Staroselskii Georgii
Daniel Frenzel