Migration to Catkin build system #75
Conversation
- deleted manifest.xml - updated CMakeLists.txt - created package.xml This patch shuold be tested. It's also neeed to change the e-mail in package.xml, it was choosen a random e-mail to make the package compile with catkin
|
Thank you very much for the patch @boris-il-forte . I think we need a branch for |
|
I think it would be easy... It's just about converting the existent cmakefile in a catkin-compliant cmakelist.txt, that's nothing special, and creating a package.xml with all the correct fields... I can help if is needed... EDIT: I made a mistake when I said "update cmake files", because the whole project is based on ARDronelib Make, create a CMakeList.txt is not as straightforward as I thought. basically I think you could choose two ways, the first is to call "make" or a script from CMakeList, and move the generated targets in the correct folders used by catkin, the second is to rewrite the entire build system using cmake, which itself should not be difficult, but definitely takes some effort. the second way is definitely the cleanest, the first is a dirty trick to force the use of Make in catkin, which is designed precisely to prevent this. |
|
👍 for a catkinized branch! I'd very much appreciate if you could maintain/transition to a catkin version of the package for hydro. I can also confirm that @boris-il-forte's changes in this PR work wonderfully for now. The installation works like before, the only command that needs to be substituted is to use
|
|
Thank you @boris-il-forte for information and @Nevik for testing. I will merge the changes/update the driver this coming week. |
|
any news about the porting? |
Parrot was supposed to release its new SDK with Flight Recorder and firmware 2.4.x support in September. My initial plan was to update the build system with the new SDK. I will wait a couple of days more for the parrot. I hope they will release their new SDK soon. |
- added rules to install ardrone_driver binary in the correct location specified by --install-space of catkin_make command
|
I realized that there were no rules to properly install ardrone_driver in the correct installation directory when you launch catkin_make - install. it should now install the driver, so that it is accessible with the command rosrun even if you specify an installation directory with the flag "- install-space" as always, needs to be tested ... |
|
simply if the node doesn't compile without is a build_depend, if it crashes at runtime (or if it's needed some other node tho work correctly) is a run_depend |
|
No news about the new parrot SDK? |
|
Ok, I managed to fix crashes on 2.4.x firmware by updating the SDK to 2.0.1. I have not tested it completely during flights, but at least it works both for 2.3.x and 2.4.x firmwares. I will patch the SDK further to support the new GPS messages. I will move on to working on The code (8c46987) is now on |
|
I just tried to catkin_make this package and came across an error. The RecordEnable.srv file in ardrone_autonomy/srv does not exist and yet it is included in the add_service_files section in CMakeLists.txt. Can anyone help me with this? Thanks |
|
Which branch are you using? if you have catkinized yourself the branch, and you don't have the .srv file, why don't you just delete it in the CMakeList.txt? it should work... |
|
@boris-il-forte I finally pushed the new driver based on SDK 2.0.1 to |
|
Ok, I'll try, and eventually fix all the problems by this evening... |
|
Ok, work done! EDIT: I've seen that is not possible anymore to update the branch of the pull request, so I've opened a new one. This shuold be closed... |
|
Follow-up on #79 |
This patch shuold be tested. It's also neeed to change the e-mail in package.xml, it was choosen a random e-mail to make the package compile with catkin
please consider to merge this patch, or to doing youreself the catkin migration: catkin packages can't depend on rosbuild packages, so all new projects based on Groovy can't use the driver anymore.