Upon starting a fresh client on a newly started server I received this weird color scheme. Also notice the extreme score.
After starting to wiggle, the wiggle would not stop anymore despite the 'stopping' text. The client eventually crashed.
I could not reproduce this bug. Next time it started fine, score was fine as well. Uninitialized variables. variables overwriting other data ?
I was just about to start or update a thread on similar problems. I had started four newly updated clients that hadn't opened this puzzle before. Three clients opened normally, one was similar what BP shows above.
The score in the my bad client was also -987,654.321. Which just happens to be 123,456.789 in reverse. That's probably not a coincidence.
I also notice that the bad client shows no density scores. As BP discovered, you'll get a crash if you try to work on this pose.
The details seem to show that the -987,654.321 doesn't relate to anything. It's likely a placeholder indicating the density calculation hasn't happened.
The good client has density scores, and everything adds up.
Here's the log file for the crash, see also the logs posted in bugs-and-feedback on Discord.
The good news is that when you restart a client with the -987,654.321 score, it opens normally in most cases.
I opened another four updated clients, and one ended up with the -987,654.321 score. So the odds of that happening seem to be 25%, based on a small sample.
Opening a shared solution didn't resolve the problem, and a quick wiggle was enough to cause a crash.
I looked at the log.txt for another -987,654.321, and noticed lines like these:
standalone.application.boinc.Puzzle: {0} starting score : nan
rho_calc 1 of 2002
rho_calc 2 of 2002
rho_calc 3 of 2002
rho_calc 4 of 2002
rho_calc 5 of 2002
rho_calc 6 of 2002
rho_calc 7 of 2002
rho_calc 8 of 2002
rho_calc 9 of 2002
rho_calc 10 of 2002
rho_calc 11 of 2002
rho_calc 12 of 2002
rho_calc 13 of 2002
rho_calc 14 of 2002
rho_calc 15 of 2002
rho_calc 16 of 2002
rho_calc 17 of 2002
rho_calc 18 of 2002
rho_calc 19 of 2002
rho_calc 20 of 2002
rho_calc 21 of 2002
…and so on….
Starting score being "not a number" seems bad.
Thank you all for reporting this, we're trying to figure out what is going on here!
Any chance you can post a screenshot after pressing Tab on a specific residue?
See the two screenshots I posted 1 November for starters. The one with the -987,654.321 score has the score in black which is unusual. BP's screenshot is also weird, not sure how you get those colors with those view options.
I'll see if I can still find the test puzzle and recreate the bad score issue.
Here a series of screenshots showing the issue, freshly captured.
Using my defaults for small molecule design, the protein looks pretty weird when this happens.
Colors are better with more common protein puzzle view options.
The value "-nan(ind)" seems to be a popular choice for the density subscore in the segment information window.
With a dark background, the colors look really weird. This one is shot on a different client, with wireframe density selected. When the density is dialed way down, a strange structure emerges. There is very little density to be seen otherwise.
As BP reported, wiggle seems to stick, and doesn't promptly respond to any stop action. A crash happens at some point, less than a minute after starting wiggle. Log file attached.
Here's a follow-on to the last image. I notice there's second weird structure in the opposite corner of the density. The two weirdos look like one weirdo that's been cut in half.
This client hasn't crashed yet, so I'll try to keep it going. Let me know if you'd like more.
The previous examples were on devprev release 26. The problem also strikes on release 27. Here's a log for a quick crash. The crash seemed to happen just after loading the puzzle, but I might have hit a stray key.
Just for comparison, in the normal case, a low threshold setting gives a nice dense cloud.
