End point lanes change, unexpectedly, when start point lanes are changed

[bdog2112]

Observed an intermittent issue that we've seen many times in the past. However, I thought it was attributed to the IDE being unstable after plugin edits, which is to be expected But, now I think it's something else and I found some steps that help to re-create it for further troubleshooting.

ISSUE: Sometimes the "end point" number of lanes changes when the user changes number of lanes on the "start point".

In the image below, one lane was added to the top RoadPoint and the middle RoadPoint unexpectedly also added a lane. The expected behavior is for lanes to be added only to the selected RoadPoint.

Steps to re-create the problem:

• Create a new scene
• Create a 2x2 road
• Select the last RoadPoint and click "+ Next RoadPoint" to add a 3rd point
• Save, close, and re-open the scene
• Select the first RoadPoint and add a lane
• The second RoadPoint should unexpectedly, also, add a lane

Not all RoadPoints exhibit the problem. But, if you limit the test case to the simple road network described, above, you should be able to observe the issue.

[bdog2112]

It's worth noting that, if you save the file pictured above, the file will show that lanes on all three RoadPoints have been changed. The third RoadPoint simply isn't redrawn to show the change. Although, if you click the third point, it will redraw and the change will become visible.

[TheDuckCow]

Thanks for logging the bug - I presumed it somehow had to do with there being two pointers to the same array index somehow. This also happens in our wheel steal demo figure 8 scene.

Let's leave it alone for now, we can bring it to another release after the launch (which I might have you prioritize working on for the couple weeks while I'm traveling, we'll set expectations in our Sunday team meeting).

[bdog2112]

When troubleshooting, earlier, I noticed that you can leave export variables null in the declaration and populate them in _init or _ready. Then, they will be considered changed from their default and their values stored in Scene files no matter what. I believe that is a solution that could work right now.

[TheDuckCow]

As far as I understand now, @bdog2112 we actually already resolved this bug right?

Closing the issue as I can't reproduce this now (and I recall we came to a solution to linked arrays), but please comment if you recall otherwise and think this is still open.

[bdog2112]

Yes. This was fixed when array initialization was moved to the _init routine.