Tested puzzle [DEVPREV] Symmetric Dimer Design
Crash on Win 10 i7 8GB RAM laptop:
Manual Shake - OK
Manual Wiggle - OK
Manual Shake - OK
Run script "Tvdl enhanced DRW 2.8.0 B" - Crash after ~ 10 sec during ShakeSidechainsSelected operation - see attached log.txt
Pretty much the same thing, Win 10, Core i7, 8 GB. Crashed during a mutate combo.
crashed for me during tweak…rotate
getting An HTTP error 0 occurred.
/portal/comment-upload/js ………..when i try to attach log file
I can reliably crash this puzzle, but the only way I have found so far is to run a script that rebuilds. I can't crash it by rebuilding by hand, or by shaking.
Using Devprev Symmetric Dimer Design puzzle:
Reset puzzle, start DRW script, all default options, crashes within 5 seconds (log copy 23)
Restart client, reset puzzle, start eDRW script, all default options, crashes within 5 seconds (log copy 24)
Try a bunch of non-rebuild stuff without crashing, then run a rebuild test script and crash in first 5 seconds (log copy 25)
Reset puzzle, run rebuild test script, crash in 5 seconds (log copy 26)
Reset puzzle, rebuild and stop after every found pose, over and over by hand, no crash. Start rebuild test script, crash immediately (log copy 27)
20170427-df30f666e0-win_x86-devprev
I made a simple test script that reliably crashes all 3 symmetry puzzles (devprev Symmetric Dimer Design, Beginner Puzzle <150 Symmetry Design, and devprev HBN) on Remix. Sometimes it is the first Remix that crashes; sometimes the 2nd. Let me know if you want log files or Windows events; I have lots. You should have lots of crash logs from all three puzzles; if not, the mechanism to send the crash logs up to you is not working.
For what it's worth, when I use a similar script that Rebuilds instead of Remixing, I see the symmetric (gray) chain change shape just before the crash. That means Rebuild finds a pose, calculates the new position for the symmetric chain, and draws it on the screen before crashing. With the Remix script, I do not see the new pose on the screen before the crash - I think it only gets drawn when the slot containing the new pose gets loaded, which is the line of code after the Remix call, and the script doesn't get that far. When it doesn't crash on the first Remix, I see the new pose from the 1st remix, just not the one that crashes.
origslot=4
rxslot=5
counter = 0
save.Quicksave(origslot)
selection.DeselectAll()
selection.SelectRange(5, 10)
while true do
counter = counter+1
save.Quickload(origslot)
print("quickload original "..counter.." succeeded")
local nfound = structure.RemixSelected(rxslot,1)
if nfound > 0 then
print("remix "..counter.." succeeded")
save.Quickload(rxslot)
print("quickload rx pose "..counter.." succeeded")
end
Have you ever thought of releasing a test version of the software (update group "test" instead of "devprev") that has extra print statements aimed at locating the symmetry bug(s) specifically? I guarantee you would find a group of players willing to install and run it.
Win7, your latest client as of today.
Crashed immediately with tvdl drw on rebuild. log (attemping to) attached.
I've the impression that several issues are linked to the hardware.
For example, due to repetitive problems on Mac Os, I installed Linux on a MacBook pro => I get the same family of problems (+ new ones).
My computers are slowing down with the years (I suspect fan dysfunctions) with more crashes. Temporary network disruption also cause crashes, as well as screen savers etc.
No crash on Mac OS
(latest DEVPREV updated today)
1) I ran my top (shared to scientist) (windows) 12246 result on a Mac (Intel Core i7, 8Go - MacBook Pro with OSX 10.8.5), with manual shake all, local shake (rebuild is disabled), Random Idealize, tvdl remix and several other recipes: no crash.
2) Ran shared to scientist 11455 pts "crash on idealize selected" result on the Mac: shake all ok, recipe with idealizeSelected ok
Conclusion: no crash on a Mac OS.
Windows 7, New DEVPREV update from today:
1) No crash (and no rebuild)on my top 12146 (end game) shared to scientist solution.
2) Ran shared to scientist 11455 pts "crash on idealize selected" result on the Mac: shake all ok, recipe with idealizeSelected CRASHED
(see log)
Thanks for the feedback everyone.
We're working on debugging this internally, but it's looking like a very nasty bug to track down.