I all,
I stumbled on an odd error on yesterday evening while finalizing my second ZIP for the "Stars of Call" Adventure Mod. It features a fixed non-linear network of 28 systems, which I'm currently refining, so I'm always testing by jumping from system to system.
At first everything seems to run smoothly. I jump from system to system, I come back and forth and I explore. Then, when I jump out of the NN 3006 system and tries to get back to it, the whole game crashes.
The error might be a "The memory cannot be read" or a "Runtime error" or "Unexepected error". The actual error seems to be quite random, or at least I couldn't get any clue as to relevant conditions.
So far, crashes only occur on this particular system. It still crashes when I jump out of NN 3006, then make several jumps and try to get back to this system.
I tried a number of possible fixes, but at first glance there's nothing odd on the system code itself and neighboring systems. I don't know for how long the problem has been there because I discovered it by trying random paths from system to system.
The problem is that it may be an app bug, a mod code bug or an odd interaction between a pushed-a-bit-too-much mod code and some unexplored corner of the app ^-^ So how to really pinpoint it ?
I'll do some extra trials. If I don't find a quick fix, I'll post the ZIP in its current state and record this problem as known bug in the readme. So people here will be warned of the problem and will be able to test and report under extra conditions, maybe finding new possible explanations.
Crash when jumping systems in Adventure Mod game
- Mutos
- Militia Lieutenant
- Posts: 218
- Joined: Thu Aug 14, 2008 3:31 am
- Location: Near Paris, France
- Contact:
Hi all,
Morning's first trials revealed some clues...
The system is just a giant asteroid field. I first fixed the number of asteroids to 5000 small and 1500 medium.
As part of the trials, I totally deleted the asteroid field, making the system bare, with only a startup marker and three jump points (stargates) markers. It didn't crash any more. So I uncommented the asteroid field and the crash reappeared.
After that, I set the numbers of asteroids to something more reasonable. It worked with 1500 & 150. So it's some kind of a memory/array/overflow/index/pointer limit on the game, that only gets visible when pushing it too much through mod code. Seems it could simply be a limit on the number of station objects in a given system, or something like that.
It's a worst-case situation, because debugging it thoroughly would involves both game and mod code... Anyway, for my part, I'll stick with 1500 & 150 asteroids and release my second SoC Mod ZIP with these ^-^
Morning's first trials revealed some clues...
The system is just a giant asteroid field. I first fixed the number of asteroids to 5000 small and 1500 medium.
As part of the trials, I totally deleted the asteroid field, making the system bare, with only a startup marker and three jump points (stargates) markers. It didn't crash any more. So I uncommented the asteroid field and the crash reappeared.
After that, I set the numbers of asteroids to something more reasonable. It worked with 1500 & 150. So it's some kind of a memory/array/overflow/index/pointer limit on the game, that only gets visible when pushing it too much through mod code. Seems it could simply be a limit on the number of station objects in a given system, or something like that.
It's a worst-case situation, because debugging it thoroughly would involves both game and mod code... Anyway, for my part, I'll stick with 1500 & 150 asteroids and release my second SoC Mod ZIP with these ^-^
- Mutos
- Militia Lieutenant
- Posts: 218
- Joined: Thu Aug 14, 2008 3:31 am
- Location: Near Paris, France
- Contact:
Hi all, hi digdug,
I'll check that, but it crashes the same with a 1GB machine and a 2GB machine. That was what prevented me from thinking of an OoM error. Indeed it seems odd that Transcendence would have set limits to object numbers.
I'll check that, but it crashes the same with a 1GB machine and a 2GB machine. That was what prevented me from thinking of an OoM error. Indeed it seems odd that Transcendence would have set limits to object numbers.