Model Following Paths is broken when shape files not kept in memory

[AlexisDrogoul]

Describe the bug
Independently from #259 , whenever I try to run Following Paths, I get the following error:

2 occurrences in 2 agents at cycle 0: vehicle_following_path0 occupies 1 lanes, and it is unable to enter road163 where there are only 0 lane(s) available
in Simulation 0
in agents vehicle_following_path0, Simulation 0
in do drive ;
in reflex commute when: current_path != nil  {
	do drive ;
}

To Reproduce
Steps to reproduce the behavior:

1. Run the model from the library

Expected behavior
No error.

Screenshots

** gama.info **
GAMA VERSION: 0.0.0 commit Not available (Git version)
COMPUTER: Mac OS X 14.6.1 on aarch64 (16 cores, 128 GB)
VIRTUAL MACHINE: OpenJDK 64-Bit Server VM Eclipse Adoptium version 21.0.3+9-LTS
AVAILABLE MEMORY: 3.5 GB / 4 GB
GRAPHICS CARD: Apple M3 Max by Apple (GL version: 2.1 Metal - 88.1)
MONITOR #0: 2056x1329 zoom 200% (primary)
MONITOR #1: 3440x1440 zoom 100%

[ptaillandier]

I was not able to reproduce this bug (GAMA git 2024 up to date, new workspace, default preferences). Did you modify something in your preference?

[AlexisDrogoul]

Indeed -- reverting my preferences to their defaults solved the issue. However, I do not know which preference may have caused that 🤔 ?

[AlexisDrogoul]

OK. I found the guilty one (but still have to understand why):

When it is false, I obtain the error. If I set it to true the error vanishes. Do you have the same problem ?

[AlexisDrogoul]

Doing further investigations, I discovered that the issue is with the attribute memoryMapped of ShapefileDataStore. When it is set to false (through the preference or directly in the code of GamaShapeFile), the model stops working. When set to true it works again.
@lesquoyb @RoiArthurB do you observe the same behavior on Windows and Linux ? Because the only information I could find is that it is better to turn it off on Windows ... (see https://gis.stackexchange.com/questions/245536/why-is-the-use-of-memory-mapped-buffers-in-shapefile-data-stores-not-recommended).

[AlexisDrogoul]

I reactivate this one as it is simply a matter of checking if the same behavior occurs on Windows and Linux ( @lesquoyb & @RoiArthurB ) -- if yes, then we will not play with this variable and leave it true all time.