Open maarzt opened 3 months ago
I could not reproduce the bug with the beta-29 on Mac. Am i missing a step?
I could not reproduce the bug with the beta-29 on Mac. Am i missing a step?
This is strange. We could reproduce this bug on Ubuntu and Windows with the beta-29. I used this dataset for testing: https://github.com/mastodon-sc/mastodon-example-data/tree/master/tgmm-mini
Do the same, but click 'Regenerate' before adding a new spot.
That would work, but users would need to be aware of this.
No I think this is a bug. Your finding confirms that it caused by the branch graph becoming seriously out of sync with the core graph. We could patch it by adding a 'regenerate' call to the remove action. What do you think?
No I think this is a bug. Your finding confirms that it caused by the branch graph becoming seriously out of sync with the core graph. We could patch it by adding a 'regenerate' call to the remove action. What do you think?
I had the same idea, but since this is different from the solutions @maarzt suggested, I would also like to have his opinion on this.
@tinevez wrote:
No I think this is a bug. Your finding confirms that it caused by the branch graph becoming seriously out of sync with the core graph. We could patch it by adding a 'regenerate' call to the remove action. What do you think?
That's a very good idea. And much simpler than what I had in mind.
To make the branch graph consistent with the model graph one could also add automatic branch graph rebuild after:
@tinevez what do you think?
Hum I am not in favor of this. Initially we had the 'regenerate' button to avoid triggering branch graph recalculation at every event. I am not sure yet how to reconcile robustness (not crashing) with not calling the regenerate routine too often.
Same issue than #226 ?
Same issue than #226 ?
Yes, they are very likely related.
I would still not close this issue, since, I think, it is good to have multiple test scenarios for a potential fix at hand.
Steps to reproduce
Ctrl + A
andShift + Del
in BDV window)a
Reason for the exception
The ModelBranchGraph tries to reference ModelGraph spots that no longer exist. Results in a call to DefaultSelectionModel.isSelected(...) with a spot poolindex -1.
Potential solutions
Preferred solution:
To make the branch graph consistent with the model graph one could also add automatic branch graph rebuild after:
Other solutions: