- pulls more of the travis-ci s3 deploy into the repo so we can potentially migrate to another CI system
- fixed the sizes output and added verbose compiler version to cmake (#6322)
- fixed filenames for firmware uploaded to s3 (was broken by the changes yesterday)
- fixed some broken git version display in cmake
- Makefile organization
- simplified .travis.yml
- added a print to know which config the nuttx patch was being applied to
- docker_run.sh now respects PX4_DOCKER_REPO for setting the docker image, but defaults to the good production nuttx image
* Recursevely go over all submodules
Using git submodules command going over all the submodules and not only the ones on the whitelist
and fixed white spaces
Hotfix for cmake configs which use include() for the module list (eg.
posix_sitl_ekf2.cmake or snapdragon). In that case the cmake parser did
not find any modules and thus the param list was empty.
The proper fix will be to parse the include() statements correctly.
- the only difference really is, that scope (the configuration.cmake) is already passed
to px_process_params via the argument --scope. The Paths in --scope are evaluated w.r.t
the path to src provided via the -s /--src-path argument.
- if no --scope is proveided. the Old scheme by simply walking the full --src-path directory
is applied
- Run in a single ssh command
- Allow to update firmware when mavlink-routerd had already been
stopped
- Be resilient to another daemon interfering the communication by
reading/writting to the UART
- Print OS, BIOS nad FPGA version
Steps to debug:
* Run gazebo (or any other sim) server and client viewers via the terminal: `make posix_sitl_default gazebo_none_ide`
* In your IDE select `px4_<mode>` target you want to debug (e.g. `px4_iris`)
* Start debug session directly from IDE
This approach significantly reduces the debug cycle time because simulator (e.g. gazebo) is always running in background and you only re-run px4 process which is very light.
Since the git hashes were being generate by cmake it would only be
generated if the header file was not present. Simple test:
$ make aerofc-v1_default
$ touch a
$ git add a
$ git commit -m tmp
$ make aerofc-v1_default
The file build_aerofc-v1_default/build_git_version.h should have the new
hashes and the correspondent .c/c.pp files should be rebuilt, but they
aren't. The end result is that checking the version with "ver git" in
the nsh console will point to the wrong commit.
This moves the generation of the header to a separate tool and enforces
the command to be executed every time.
Janis Dzerve