fabian Posted February 13, 2020 On 2/12/2020 at 12:21 AM, Finny said: , and crispy just says fuck you and closes without warning. Which dehacked patch was that? 0 Share this post Link to post
Krull Posted February 13, 2020 (edited) . Edited September 2, 2023 by Krull 0 Share this post Link to post
Finny Posted February 13, 2020 (edited) 14 hours ago, fabian said: Which dehacked patch was that? the file i posted on the last page will eventually crash that way if you keep playing around with paladins (shield & rocket launcher enemies) for long enough, but i threw together something that should get you the crashes pretty quickly & consistently if you want to see for yourself. get it here. that's a much more straightforward test where lost souls are modified to go straight to Stop after SkullAttack. the map is a mix of lost souls and imps, and the lost souls are grounded - seems much more likely to induce the crash this way. you should be able to run around the mob with your weapon of choice and in a few seconds you'll get the crash, where crispy just exits and prboom+ spits out a signal 11 error. while you're here, i was also wondering if this was a bug or just dehacked funkiness: when you set a weapon that fires something, hitscan or projectile, to use infinite ammo, it ends up using total shotgun ammo instead, like your pool of shotgun ammo depletes. i found that manually setting the ammo type from 4 (what whacked uses for the infinite ammo setting) to 7 will use each type of ammo's total amount. this is another thing that happens in both crispy and prboom+, so i was assuming this was normal behavior that some port authors specifically adjusted. 8 hours ago, Krull said: How long did you have to wait and how many projectiles & lost souls did it take before it crashed? I just tried the nuts map stress test for ~5 mins in prboom & glboom (both the last complete versions and the newer test version) and nothing happened. Also, were you using 0 duration Stop frames? newer test version? i'm assuming you've whipped up some test stuff of your own because i haven't uploaded anything since the 10th until now... anyway, the file linked above is pretty much guaranteed to crash pretty quickly, and uses a 0 duration on SkullAttack before the Stop call. you can increase that and it'll make it take longer to crash, but it should still eventually crash provided you don't give it a ridiculously large duration to where the lost souls can't call Stop before running into something anyway. i tested with a 4 frame duration and still eventually got a crash. initially i thought adjusting the duration of the Stop frame itself didn't affect anything - the actor would immediately return to the Spawn state without even waiting the entire duration of the Stop frame anyway - but it seems setting the Stop frame to something extremely high (i tried 100 frames) is somehow preventing crashes, even though in testing i don't see any notable visual difference between having that set to 4 frames, which is what its set to in the file above. 0 frames seems to make the crash darn near instant. Edited February 13, 2020 by Finny 0 Share this post Link to post
Krull Posted February 14, 2020 (edited) . Edited September 2, 2023 by Krull 0 Share this post Link to post
fabian Posted February 16, 2020 On 2/13/2020 at 11:55 PM, Finny said: the file i posted on the last page will eventually crash that way if you keep playing around with paladins (shield & rocket launcher enemies) for long enough, but i threw together something that should get you the crashes pretty quickly & consistently if you want to see for yourself. get it here. Thank you! Interestingly, this appears to trigger a spechits[] underflow. The culprit seems to be more in this part of the BEX patch rather than in modifying the SkulAttack state to go to Stop. Bits = SHOOTABLE+SOLID 1 Share this post Link to post