Errors attempting Nav2 demo using TurtleBot3

[threent2]

1 After changing the base module to the gui variant I get an error as Gazebo attempts to open from the launch file. The relevant lines from the log file are :-

[gazebo-5] libEGL warning: Not allowed to force software rendering when API explicitly selects a hardware device.

[gazebo-5] Segmentation fault (Address not mapped to object [0x8])

This is easily fixed by unsetting LIBGL_ALWAYS_SOFTWARE. I did have this happen once before but did not raise it as an issue as it affects so few people. I do wonder if there was a better long term solution to my graphics problem but it is not worth spending time on.

2 Rviz2 opens with map as the fixed frame rather than base_link. This can be changed manually. Also TF is not loaded but can be loaded manually. This still leaves an error in the Map section - "No map received" I am not sure how to fix this.
I will see if there is a config file or maybe parameters set at load time to fix the other two inconveniences.

PS I don't know if you spotted that I added the solution to Issue 2 - sine_ctrl. If you build the module by following the text in the book it works fine. The module in Github is structured differently and has extra text.

[threent2]

1. Trying to save and then re-load the rviz configuration causes rviz to abort. I will just make each change manually.

2. Although rviz shows "No map received" these lines from the log suggest a map has been loaded : -

[rviz2-7] [INFO] [1759054458.301709089] [rviz]: Subscribing to: /local_costmap/voxel_marked_cloud
[rviz2-7]
[rviz2-7] [INFO] [1759054458.324612344] [rviz]: Subscribing to: /global_costmap/voxel_marked_cloud
[rviz2-7]
[rviz2-7] [INFO] [1759054458.341969482] [rviz]: Subscribing to: /mobile_base/sensors/bumper_pointcloud
[rviz2-7]
[rviz2-7] [INFO] [1759054458.345925967] [rviz]: Subscribing to: /local_costmap/voxel_marked_cloud
[rviz2-7]
[rviz2-7] [INFO] [1759054458.352580788] [rviz]: Subscribing to: /global_costmap/voxel_marked_cloud

I will attempt to continue following the text and see if I can proceed or if more errors occur.

[threent2]

It seems I am just impatient. If I ignore the initial errors and just continue to follow the text and use "2D pose estimate" to set the initial pose then everything corrects itself in Rviz (picks up correct map) and I can set pose and waypoints.
At the moment I still seem to have a problem with the robot model (which is not visible) which does not show as active. If I manually activate it I get an error under URDF - Errors loading geom...
Update
Although the model does not show in rviz it does show in Gazebo (if I zoom in to it).
Navigation seems to work if I pick destinations that I can see should be possible but just aborts after a short time if there is an obstacle in the way.
Second update
I have been unable to get either Docker Compose or VSCode working. As both give similar errors I will just give my experience of Docker Compose here.
As background when just using Docker I have to set the Initial pose before everything sets up cleanly. Initially the Global Status shows an error but this corrects itself when the 2d Pose estimate is set.

When attempting to start from Docker Compose (having done all the extra steps when attaching to the container - including xhost +) It goes in to an infinite loop showing :-
Timed out waiting for transform from base_link to odom to become available, tf error: Invalid frame ID "odom" passed to canTransform argument target_frame - frame does not exist
and I never get either a Gazebo or Rviz2 window to set the initial pose.
I tried to amend the tb3_simulation_launch.py so that it didn't try to load rviz by removing
ld.add_action(rviz_cmd)
but I still get the same error