Closed mlesnick closed 7 years ago
Original comment by Bryn Keller (GitHub: xoltar):
I can reliably reproduce, and I have a lead. Hopefully I can fix this today.
Original comment by M L (GitHub: mlesnick):
Strange. It is quite repeatable on my system, even after restarting my computer. It will be interesting to see if it happens on Bryn's MacOX.
To get a sense of how new the bug is that causes this, I checked out commit 35302ae from 2:38 a.m. Dec. 28th. H0 10x10 does not crash RIVET for me in this commit. So that suggests that whatever this bug is, it was introduced recently.
Original comment by Matthew Wright (GitHub: mlwright84):
I have not been able to duplicate this problem. I just ran the cases that Mike mentioned, using the code in commit ee35379, and it worked fine.
Original comment by M L (GitHub: mlesnick):
To make RIVET hang, compute H0, and try moving the line diagonally upward from the starting position, without changing the slope, as in the picture.
I was also able to make an error like this happen with H0-15x15 bins, and with H1-10x10bins, for the same data file. On a different data file, I did not see any such error.
Originally reported by: M L (GitHub: mlesnick)
I played a bit with Matthew's Betti fix, on the 240 point example. I had no problems.
Now I am trying Bryn's commit fa89ec4. The problem about the file formats looks to be fixed.
However, for the 240 point example, 10x10 bins, as soon as I move the slice line, RIVET hangs indefinitely, giving me an OSX spinning pinwheel.