Open frs69wq opened 8 years ago
It looks like a problem from the N-body barnes-hut graph positioning algorithm. It is very hard to replicate this kind of bug, I thought I have fixed most of them. The problem is that my implementation in libtupi is not very robust (works nicely when nodes are not added/removed during the algorithm execution). Since I keep a thread to execute this algorithm, it may be a concurrency problem (I have to stop the thread to add/remove nodes). The files you have used as input are the exact ones generated by the simgrid example you mention?
Hi,
I was playing with viva on one of the traces produced by SimGrid (in examples/msg/trace-masterworker, see the tesh file). I launched viva like that:
viva trace-masterworker.trace trace-masterworker.cat.plist
I have the 3 windows (graph, timeline, and config) opened and can play with the simulation visualization (even though I don't fully understand what I see ;)). However, if I try to untick any of the items (but the "toplogy" ones) in the Name / Nature window, viva crashes with this error message TupiException at dict_insert_element (/.../viva/src/libtupi/dict.c:63): [element is already present]
any hint on that?
Best F.