Crash playing sci2.wad MAP13 (Doom 2)
When playing sci2.wad under Doomsday 1.14.1, Doomsday crashes with a Segmentation Violation on MAP13 when you turn right after going through the blue door. I tested this under Doomsday 1.13.2 and it does not crash. I've attached a zip file containing the wad file as well as my user folder (savegame, config etc).
#1 Updated by danij almost 9 years ago
I've tried to reproduce this on Windows but so far I've had no luck. Can you provide any more info? Are you loading any other addons?
Edit: I've now replicated the problem. Seemingly it is dependent on one playing through the whole map, so presumably there is some special being triggered which leads to the P_Callbackp error.
#2 Updated by danij almost 9 years ago
- Category set to Defect
#3 Updated by danij almost 9 years ago
- Status changed from New to In Progress
- Assignee set to danij
#4 Updated by danij almost 9 years ago
- Target version set to 1.14.2
- % Done changed from 0 to 100
#5 Updated by danij almost 9 years ago
- Status changed from In Progress to Closed
#6 Updated by skyjake almost 9 years ago
- Status changed from Closed to Resolved
#7 Updated by skyjake almost 9 years ago
Marked as Resolved until cherry-picked.
#8 Updated by danij almost 9 years ago
- Status changed from Resolved to Closed
#9 Updated by danij almost 9 years ago
- Subject changed from doomsday 1.14.1 Segmentation Violation (sci2.wad) to Crash playing sci2.wad MAP13 (Doom 2)
Fixed|World|SectorCluster: Crash with sci2.wad (Doom2) upon triggering the exit crusher
The crash was the result of an incorrect assumption that all sector
clusters will have at least one outer boundary edge. While this is
logically true there are some special cases in which no boundary is
recorded (e.g., the one unique outer halfedge is twined with an egde
originating from a degenerate BSP leaf). In this case, the trigger
is a so-called "control sector" outside the map attempting to update
surface light decorations when it's ceiling plane moved.
Todo for later: There is no need for such a cluster to be observing
plane movement at all. To address this properly we need two things;
1) BSP building algorithm that guarantees topologically sound and
geometrically correct data, 2) more intelligent SectorCluster.