Linux64 version crashes with new script

Started by BitSpawn

BitSpawn Lv 1

Hello
Every time I write a new script, or edit an old one, Foldit crashes with a segment violation with no log.
When I run Foldit again, the script loads ok.
It seems that the script doesn't load in memory, and it's necessary to load from all.macros at the beginning.

version-binary.txt: 9f3ed4b5c87641d6587198ebc1e40757
version-database.txt: 307d41cd20e292f4c5d928df7ed9e08b
version-resources.txt: b3aafeebbef6dea92a55ccea0c0658e4
Linux version 2.6.34.7-0.5
Linux Dist: OpenSuse 11.3
Foldit[15065]: segfault at 40699f96 ip 0000000040699f96 sp 00000000ffbfb9cc error 14
Foldit[19176]: segfault at c007b80e ip 00000000c007b80e sp 00000000ffa431bc error 14
Foldit[19408]: segfault at 90dd410 ip 00000000f5992557 sp 00000000ffed05e4 error 4 in game_library.so[f4b13000+27e8000]

Each error in different moments with different scripts, the segfault is not always in game_library.so

BitSpawn Lv 1

Scene 1

  • You modify the script
  • Save
  • Run it inside the editor window
  • Run properly

Scene 2

  • You modify the script
  • Save
  • Close the editor Window
  • Open coockbook and run the script
  • Crash

Last minute! :)
My all.macro is > 3MB
I have deleted scripts, and with 1.3MB all run properly :)

BitSpawn Lv 1

With a small all.macro it sometimes fails, yet.
A way to avoid the crash is don't run the script after edition. If you do another action before (restore best, or save and load) and then you run the script, it doesn't crash.

BitSpawn Lv 1

Verified previous message: never hangs with an action before running recipe.
I have also seen that slowing down the execution no errors appear.

Searching something wrong I've found what looks like persistent memory leaks errors, mainly in method chemical::ResidueType::clone() and creating the object ObjexxFCL::FArray3D::dimension(...), both of them called from load_rosetta() function. I hope it's useful