Some shares load lower on 851

Started by smilingone

bkoep Staff Lv 1

Just to be clear, what took place between the 'before' and 'after' script outputs? You just saved and reloaded the solution? On the same client or a different client?

smilingone Lv 1

Script stopped. Restored to credit high and ran script and saved output as before file. Saved solution and closed client. Reopened same client and solution loaded lower. Ran the after file at this point.

bkoep Staff Lv 1

The most recent devprev update includes a fix that we think can "circumvent" this bug, although we still haven't identified the root of the problem. We'd really appreciate if some of you could try out the devprev puzzle, <a "http://fold.it/portal/node/997310">Broken Symmetry Loads</a>, and let us know if you continue see inconsistent scores in your loaded solutions. Thanks!

bkoep Staff Lv 1

Thanks for collecting this data, smilingone—it gave us some ideas on what might be going on, although we're still trying to figure this one out. If you get a chance to try the new devprev puzzle, let us know if you still have problems saving and loading.

brow42 Lv 1

I suppose…we could write a LUA script that saves all the score parts, # of rotamers, and maybe something like the distance from the last atom to C-a, and stuffs this all into a note. It could then do the same thing on the loaded puzzle and compute the difference for you. Would this be useful? And not a problem (you don't want a giant note displayed!). Although, it looks like it's already clear that the atom positions are not being restored accurately, and the workaround might make the effect disappear in the next patch.

LociOiling Lv 1

I encountered a crash running the Broken Symmetry puzzle (aka the B.S. puzzle). The crash was "asymmetry introduced by ActionIdealize". This makes sense, as I was running an idealization script at the time. I saw similar problems during the earlier trimer puzzles. On those puzzles, I saw an "unknown atom" crash at least a couple of times while idealizing.

After the crash, the autosave for my high score loaded two points lower than its stated value. I've shared the autosave with the scientists.

jflat06 Staff Lv 1

Thanks for the feedback - I'm looking at your crash report right now.

Do you know how long the script had been running when it crashed?

LociOiling Lv 1

Used upload for scientists to share a BS solution which shows a score of 14031.744, but loads as 14030.900. The difference is 99% in bonding subscores of the last two segments, 64-65.

LociOiling Lv 1

The script started 3/24/14 23:24:23 CDT, entered a new cycle 03/25/14 01:01:57 CDT, and crashed shortly thereafter, around 1:03 CDT. So just under 40 minutes total run time. I archived the whole directory after the crash, can share whatever you need.