1533 crashing on ideal SS

Case number:845829-2005398
Topic:Crash/Hang
Opened by:Bruno Kestemont
Status:Open
Type:Bug
Opened on:Wednesday, June 13, 2018 - 07:15
Last modified:Sunday, June 17, 2018 - 23:27

Crashing on Linux and Windows when running Ideal SS on end segments.

(Wed, 06/13/2018 - 07:15  |  14 comments)


Joined: 09/24/2012
Groups: Go Science

Debug file in annex

Joined: 09/24/2012
Groups: Go Science

ERROR: asymmetry introduced by ActionIdealizeSS

and the full log in annexe

dbuske's picture
User offline. Last seen 2 weeks 2 days ago. Offline
Joined: 09/22/2011
Groups: Beta Folders

Same here. It doesn't crash everytime I use ideal SS.
It might do the first one ok then crash the second time.
Windows 10 preview
Intel duo-core.

Joined: 09/24/2012
Groups: Go Science

Indeed, for me SS crashes 30% of the times (on end segments). I can repeat it and share it to the team if needed.

bkoep's picture
User offline. Last seen 5 hours 22 min ago. Offline
Joined: 11/15/2012
Groups: None

Yes, please share!

We're still having a hard time tracking the source of this bug. If you have a set of simple steps to reproduce the crash (even 30% of the time), that would be really helpful—especially if there is just one or two steps involved!

Joined: 09/24/2012
Groups: Go Science

Shared to scienstists with the name "crashing repeat", score 0 and description "do ideal SS on the latest helix and it will crash".

LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

Here are the steps to reproduce:

1. Start Foldit
2. Load Puzzle 1533
3. Reset puzzle
4. Use selection interface
5. Select 2-25
6. Mark as helix (l, k)
7. Idealize SS (5)
8. Select 29-52
9. Mark as helix (l,k)
10. Idealize SS (5)
11. Select 56-79
12. Mark as helix (l,k)
13. Idealize SS (5).

The client crashes on step 13. Log.txt and debug.txt attached.

I did these steps manually.

Alternately, do steps 1-3. Open a blank recipe (via recipe editor -> new (ScriptV2)) and insert this code:

selection.SelectRange ( 2, 25 )
selection.SelectRange ( 29, 52 )
selection.SelectRange ( 56, 79 )

Run the recipe. Then, manually mark the selected segments as helix (l,k) and idealize SS (5). The client crashes.

A little slower, but the following version also works:

selection.SelectRange ( 2, 25 )
selection.SelectRange ( 29, 52 )
selection.SelectRange ( 56, 79 )

for ii = 1, structure.GetCount () do
    if selection.IsSelected ( ii ) then
        structure.SetSecondaryStructure ( ii, "H" )
    end
end

Again, run the recipe. The client crashes when you manually idealize SS (5).

I also shared two solutions, "step0" and "step3". "step0" has first helix built, crashes reliably when you build the second and idealize, then build the third and idealize (basically, start at step 8 above). You can also use either recipe fragment to produce the crash.

"step3" has the first two helixes complete and the third marked. It sometimes crashes when you select and idealize the third. Loading the solution, then restarting the client seems to work best, but it's hit and miss. Using either version of the recipe seems to make it a reliable crash, however.

Just for reference, here's the structure in print protein/SS edit format and segment ranges:

LHHHHHHHHHHHHHHHHHHHHHHHHLLLHHHHHHHHHHHHHHHHHHHHHHHHLLLHHHHHHHHHHHHHHHHHHHHHHHHL

2-25
29-52
56-79
LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

Alternate 3, for Lua fans.

1. Start Foldit
2. Load Puzzle 1533
3. Reset puzzle
4. Use selection interface
5. Open recipe editor, select "new (ScriptV2)"
6. Insert the code below
7. Run the recipe.
8. Close the recipe editor.
9. Do Idealize SS via keyboard shortcut "5".

The client crashes.

rtable = {
    {  2, 25, }, 
    { 29, 52, },
    { 56, 79, },
}

selection.DeselectAll ()

for rr = 1, #rtable do
    selection.SelectRange ( rtable [ rr ] [ 1 ], rtable [ rr ] [ 2 ] )
    for ii =  rtable [ rr ] [ 1 ], rtable [ rr ] [ 2 ] do
       structure.SetSecondaryStructure ( ii, "H" )
    end
end
LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

It seems like having only some segments selected is key.

If you do alternate 3 above, but then select all (ctrl + a) before Idealize SS, all is well.

But even if you manually select the loops between the helixes, leaving only 1 and 80 deselected, Idealize SS still causes a crash.

Edit: after running the alternate 3 recipe, manually add 1 and 80 to the selection, but leave the loops between the helixes unselected. Idealize SS does not crash.

LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

Pure play, no recipe:

1. Start Foldit
2. Load Puzzle 1533
3. Reset puzzle
4. Use selection interface
5. Select 2-79.
6. Mark as helix (l, k)
7. Idealize SS (5).

Respawn time again.

bkoep's picture
User offline. Last seen 5 hours 22 min ago. Offline
Joined: 11/15/2012
Groups: None

Bingo! This is exactly what we've been looking for! Minimal steps that lead to 100% crash rate. This should lead us straight to the source of the asymmetry bug.

Great work—thanks LociOiling!

Joined: 10/30/2012
Groups: Beta Folders

{
"build_id" : "20180312-1714f56737-win_x86"
"crash_id" : "f51be21e-060e-441a-8172-5eed402abcc9"
"current_actions" : ""
"current_tool" : ""
"file" : "utility/keys/SmallKeyVector.hh:545"
"machine_id" : "f05f6012-674f-4b6b-9c58-5225ecc72be7"
"macro_id" : "-1"
"message" : "Assertion `active( key )` failed."
"player_id" : "0"
"puzzle_id" : "0"
"puzzle_running_time" : "638659"
"running_time" : "638659"
"timestamp" : "1521730272"

LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

This one may be different, but crashed running an idealize recipe.

Terminated with "ERROR: asymmetry introduced by ActionGlobalMinimize".

Joined: 09/24/2012
Groups: Go Science

on 1530 as well: trimer problem
https://fold.it/portal/node/2005387

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons