Conversation
|
I think this crash generation could be useful for more than just the SUMO crash. |
|
Running the docker instructions, the |
|
I tried to redo it with the current HEAD of sumo but get a different crash: Is this to be expected? |
|
@behrisch Yes, this is very possible. It is likely that SUMO's behaviour has changed between the version I think I have found a way to more quickly generate the crash I will try to get a new container up for tomorrow. |
|
@behrisch I have created a new docker image and updated the instructions. |
|
This should eventually be converted into a proper PR for more general bug-testing purposes. |
|
I will work to create a proper utility out of this. It was useful to hunt down the bug with SUMO but should also be equally useful in doing the same with most other bugs and for testing determinism. |
|
It is not even certain that this is an issue any longer but it is necessary to confirm it to merge #1884.
Edit: moved to #1842 |
This is a branch that contains scripts for generating a crash and reproducing it.
Reproduction:
Option 1 - Docker
If the crash runs into
assert action == base_actionafter switching out dependencies then the crash will need to be regenerated:Option 2 - Repo
External dependencies:
Ubuntu 18.04
Python3.7.5
Eclipse SUMO 1.8.0
Generating a new crash:
Expected output