Closed servoz closed 2 years ago
In fact, Mia crashes with the V2 controller. With the V1 controller display, Mia does not crash ... but the output remains at 0.0. On the other hand soma_workflow indicates that the right result has been obtained:
Oh I see! In fact we just have to click on the brick, after the execution, to see the right output result... I think in this case it is necessary to change the codes to make an update of the controller which does not seem to take place with the V1 controller. There is still the case of the V2 controller which makes Mia crash in this case of process.
The above error message with segmentation fault was already pointed out by @denisri on the #11 ticket of mia_processes!
Yes it's the problem I faced when trying populse/mia_processes#11, and I could not understand what's going on (I have spent like 2 days fighting with it). The crash happens:
So it really looks like a threading concurrent access problem: I suspect the value modified during the run triggers a callback somewhere, which updates the GUI from the wrong thread. But even after looking a while I could not identify it.
Oh my god, this is not far from being the most complicated type of issue to fix ...
I think I've got it ! If you can check when you have time...
Yes, you did it!!! The GUI V2 controller no longer crashes Mia. That's wonderful! I'll look to refresh the V1 GUI at the end of the run.
A fake brick was created to test the execution step of a minimal process (initially related to the #11 ticket in mia_processes). This process contains just an instantiation of traits in the constructor, nothing special in list_outputs() and only an addition in run_process_mia().. At runtime, Mia crashes with a segmentation fault: