Hard Light Productions Forums
Modding, Mission Design, and Coding => FS2 Open Coding - The Source Code Project (SCP) => Topic started by: General Battuta on October 26, 2015, 10:13:05 am
-
Tester's hung up on this weird-ass issue he keeps getting on the October 24th AVX nightly. Repros every time at the same point. Seems to be maybe around a ship arrival?
http://pastebin.com/NwFKFwLX
What is this? Never seen it...
ASSERTION: "!(delta_time < 0.25f && vm_vec_dot(&objp->orient.vec.fvec, &tvec) < 0.1f)" at aicode.cpp:1240
A ship rotated too far. Offending vessel is Sanctuary Alpha 2, please investigate.
Int3(): From c:\code\fs2_open_0\code\globalincs\windebug.cpp at line 1048
-
We've got a workaround but I'm curious what the hell this is and why it crashes even on release...
-
It's not in the Oct 21 build? Between Oct 6 and Oct 21 the build environment moved to my new PC, should be the same configuration as my previous rig but there's a chance something is not quite right with the IDE setup. However if it was really not in the Oct 21 build but in the Oct 24 build, it would almost definitely have to be narrowed down to one of the changes in the log then.
-
Definitely the oct 24. Frig is gonna try rolling back a few builds to see if it goes away.
-
Frig reports:
Tested 61f2e78 and 76b6f95, worked fine.
The issue comes from the latest nightly AVX build.
-
AVX? That'd explain why I didn't get it (running a standard SSE2 build). And I though my rank.tbl crash was the most bizzare one we've had... :)
-
Can try the latest nightly to see if it was either a resolved build environment or code issue. I'll have these added to Task Scheduler soon enough so I don't have to keep remembering to run the script from work.
http://www.hard-light.net/forums/index.php?topic=90717
-
We've got a workaround but I'm curious what the hell this is and why it crashes even on release...
(emphasis mine)
If it's crashing on release, the debug assertion being triggered here is not helpful (because the thing it's complaining about shouldn't cause a crash in release code—the debug code appears to have been added by :v: to track down a physics oddity, not avoid a crash).
Is it at all possible your tester could run the Release build through Visual Studio's debugger so we can find out where it's crashing?
-
I just hit this Assertion. Still not sure it's related to a crash, but it was with fs2_open_3_7_3_SSE_20151203_aaa80df-DEBUG.exe. I was running the Media VPs 2014, and the mission Massive Battle (Beams). I also happen to have the debugger still running and attached, although I'm not sure for how long I can maintain that. I am willing to gather any info requested though. Might even be persuaded to let a dev VNC in if it would be useful. The :v: comments seem to indicate a way to gather more info about what led to the assertion, so I was tempted to try those out.