Rebuild lanelet2-io with correct pugixml and ros-humble-ros-workspace with fastrtps fix

[traversaro]

@wep21 noticed that after the build in #242, the build ros-humble-lanelet2-io package uses pugixml 1.15, while rviz uses pugixml 1.14 . This was a consequence of us not using the conda-forge complete pinning file due to prefix-dev/rattler-build#1285, and a recent pugixml update conda-forge/pugixml-feedstock#27 .

As only the lanelet2-io package is affected, the best solution is to just rebuild this package with the correct pugixml version, to permit these I done the modification in vinca proposed in RoboStack/vinca#67 .

[traversaro]

The PRs are rebuilding a bit too many packages, probably there is something wrong in RoboStack/vinca#67 .

[traversaro]

The PRs are rebuilding a bit too many packages, probably there is something wrong in RoboStack/vinca#67 .

No, the problem is present also in other PRs that do not change anything else. I guess the problem is that there were new releases (ros/rosdistro#44186 and ros/rosdistro#44185) and somehow vinca is not skipping them as the version is new, even if the build number is the same?

[traversaro]

The PRs are rebuilding a bit too many packages, probably there is something wrong in RoboStack/vinca#67 .

No, the problem is present also in other PRs that do not change anything else. I guess the problem is that there were new releases (ros/rosdistro#44186 and ros/rosdistro#44185) and somehow vinca is not skipping them as the version is new, even if the build number is the same?

I can't reproduce the issue locally, it is like for some reason locally trigger_new_versions is false, while in CI is true , see https://github.com/RoboStack/vinca/blob/aa102b280adcad0374e4d231b25d21bf0db27d61/vinca/main.py#L987 .

[traversaro]

The PRs are rebuilding a bit too many packages, probably there is something wrong in RoboStack/vinca#67 .

No, the problem is present also in other PRs that do not change anything else. I guess the problem is that there were new releases (ros/rosdistro#44186 and ros/rosdistro#44185) and somehow vinca is not skipping them as the version is new, even if the build number is the same?

I can't reproduce the issue locally, it is like for some reason locally trigger_new_versions is false, while in CI is true , see https://github.com/RoboStack/vinca/blob/aa102b280adcad0374e4d231b25d21bf0db27d61/vinca/main.py#L987 .

I found the problem. For some reason since oursland/ros-jazzy@f916d0e we are using the -n (that is the command line option for trigger_new_versions) only in the PR CI. I think we can safely remove it.

[Tobias-Fischer]

LGTM, thanks - only nit pick: could you please change "Remainder" to "Reminder" in the vinca*.yaml files?

[traversaro]

LGTM, thanks - only nit pick: could you please change "Remainder" to "Reminder" in the vinca*.yaml files?

Done in 3f01400 .

[Tobias-Fischer]

Cool - if you mark it as ready to review, I'll merge :)

[traversaro]

Cool - if you mark it as ready to review, I'll merge :)

The non-PR CI jobs use the master branch of vinca, so we need to first merge RoboStack/vinca#67 or change the CI to point to the PR branch (that I would avoid).

[traversaro]

I also bumped the build bumber of ros_workspace as it needed to solve prefix-dev/pixi#2910, the downside is that now we also need to merge RoboStack/vinca#68 before merging this PR.

fyi @ruben-arts

[Tobias-Fischer]

I think we can merge now?

[traversaro]

I think we can merge now?

Sorry, I had to refresh pixi.lock with the latest vinca commit. Now it is ready to go (if CI is happy).

[traversaro]

@wep21 can you check if this fixed your problem?

[wep21]

@traversaro @Tobias-Fischer The issue seems to be solved. Thanks!