The 3.3.0 release thread

[d_m]

I've also run into this; I didn't have the foresight to preserve the save at the time, but I do recall that it happened just as I landed the last blow on Smeagol - maybe an issue with his item drops?

Hmmm, that's interesting. I wonder if there is a code path for drops that somehow fails to initialize an object?

Thanks for the info!

[Derakon]

Twice in the past few months I've gotten a bus error when writing a comment. I don't know if that's enough to go on, but something screwy's in there somewhere...

[scud]

Can you remember anything about what you were doing prior to this? Do you have a panic save?

No panic save, and I don't think I'd just run over a drop.

I was waving and calling 'coo-eee!' at a nexus vortex that had recently swapped me about in an unimaginably cruel fashion, in the hope that it would reconsider.

I have to say this isn't something I've done before...

***

Edit: I was in a corridor; there were half a dozen Ts between me and the nexus vortex. I was standing still and smiting the onrushing Ts so I guess there would have been drops, but I'm absolutely certain I didn't move over one. Although thinking about it the 'drop creation' rather than the 'pick up' is probably the significant part, right?

[d_m]

No panic save, and I don't think I'd just run over a drop.

I was waving and calling 'coo-eee!' at a nexus vortex that had recently swapped me about in an unimaginably cruel fashion, in the hope that it would reconsider.

I have to say this isn't something I've done before...

If I understand correctly, the error condition isn't that something necessarily just dropped, just that a dungeon grid the game was interested in seemed to have an invalid object in it.

Maybe the object just got dropped, maybe it was fine and got messed up somehow, maybe it magically popped into existence. Who knows?

[scud]

The most important thing is that:

a) I'm back on the staircase

b) I can just walk back up those stairs with 40-something SP rather than 2 SP

[d_m]

Twice in the past few months I've gotten a bus error when writing a comment. I don't know if that's enough to go on, but something screwy's in there somewhere...

Hmmm, ok. I will take a look at that code. It's probably my fault

[Magnate]

If I understand correctly, the error condition isn't that something necessarily just dropped, just that a dungeon grid the game was interested in seemed to have an invalid object in it.

Maybe the object just got dropped, maybe it was fine and got messed up somehow, maybe it magically popped into existence. Who knows?

This sounds like the same bug that someone else had when they killed Ar-Pharazon, yet it wasn't reproducible when re-loading from save and killing him again. So it can't be related to drop *generation*, because his drops were generated before the save (when he was created). So it is indeed something to do with objects and the floor - maybe there's a bug in floor_carry? Or maybe not ...

d_m: I think this might be my fault. When working on level feelings I refactored object generation so that the feeling is adjusted in the calling function, not in the obj-make code itself. One of the changes was to change make_object from a bool (yes an object was created) to an s32b (the value of the created object). So wherever there used to be if (!make_object...) .... I replaced that with make_object(...); if (!o_ptr->kind) ... do you think you could give those changes a second opinion? They're in 40d63b8.

[Magnate]

In my games, I haven't changed equipment choices at all. Cover resists as best I can, and then choose the best AC combination that does not result in slowing. The *only* difference is that monsters hit me less, and when they hit I assume [I haven't looked] less damage gets through to reduce my hp. I consider that to be nerfing monster melee, although perhaps there is better term to use.

I had a recent warrior win in under 250K turns. I don't remember even getting close to that in 3.0 or 3.1. I think the reduction in damage from monster attacks played a large part.

You should not have been taking any less damage, since the percentage reduction in physical damage is the same as it's always been - capped at 60%. The change is that 240AC is required to hit the cap instead of 150. If you're saying that 240AC is significantly easier to come by than 150 used to be, then the scale isn't quite right. But if you're saying that monsters not hitting you is making a significant difference, than that could be right (as I noted above).

Have you ever worn Celeborn, and thought "I'd use something else if it had 20 less AC"?
Have you ever not worn Celeborn and thought "I'd use Celeborn if it only had 20 more AC"?
The AC doesn't affect your choice. The only thing boosting its AC does is to make the game easier.

Well, I know that several people have posted saying that the new AC values *have* made AC a factor in making their equipment decisions. Thraalbeast says in this very thread that he chose kit to optimise AC at the expense of other attributes. If you are not considering AC at all in your decisions and yet still getting benefits from higher AC, then yes something is not quite balanced. We need to work out whether it's the damage reduction or the increased misses, and adjust accordingly. (The damage reduction scale I can easily tweak for 3.3; I don't want to mess with to-hit until later when the more serious combat changes come in.)

[TJS]

Are the pac man level types going into the next release?

I don't really like them much, but I do like the cave type levels.

[Thraalbee]

Providing that it is still uncommon to get this high AC, I won't say AC is broken. At least not until having seen the effect it has on the top unique's and Morgoth.
...

Maybe AC is a bit too effective. Just defeated Atlas who can hit really hard as you know (13d13). I fought 100% melee with an ok but not great weapon but had no problems at all, only lost less than 200 HP in total.

Ranger cLvl 49.
Started fight buffed up (shield, speed, prot evil, heroism, berserk),
AC: 304
Speed: +31
HP: 1015 --> 853 after fight, no heals used
Melee damage: Avg 453 per turn .vs. giants

R01: Fear hits 3 of 4,      Atlas hits 1 of 4
R02: Fear hits 7 of 9,      Atlas hits 0 of 4
R03: Fear hits 6 of 8,      Atlas hits 1 of 4
R04: Fear hits 4 of 8,      Atlas hits 1 of 4
R05: Fear hits 5 of 8,      Atlas hits 0 of 4
R06: Fear hits 4 of 9,      Atlas hurls a boulder, miss
R07: Fear hits 7 of 8,      Atlas hits 1 of 3, cave collapse, fear dodges
R08: Fear hits 7 of 8,      Atlas hits 1 of 4
R09: Fear hits 5 of 7,      Atlas hits 1 of 4
R10: Fear hits 4 of 8,      Atlas hits 0 of 4
R11: Fear hits 6 of 8,      Atlas flees in terror
R12: You have slain Atlas, the Titan

/Mathias

[jens]

We need to work out whether it's the damage reduction or the increased misses, and adjust accordingly. (The damage reduction scale I can easily tweak for 3.3; I don't want to mess with to-hit until later when the more serious combat changes come in.)

Well it seems quite obvious that it's increased misses that is the problem. Both from the discussion on this thread, and from your comment earlier that you scaled the damage reduction but not the to_hit.

I very much do not agree with you statement not to mess with the to_hit until the big redesign of combat. There are a lot of that going around, i.e. 'there is a big change coming for X, so there is no point in balancing X right now'. That leaves us with an unbalanced game right now, that will get big changes in a few years. But if you keep the same thought about not balancing things that will change in the foreseable future, we will never have a balanced game.

Now, I don't really know the formulas involved in this particular case, but if someone points me in the right direction, or posts the formulas involved, lets see if there isn't a minor tweak we can do to improve things right now.

[Jazerus]

This sounds like the same bug that someone else had when they killed Ar-Pharazon, yet it wasn't reproducible when re-loading from save and killing him again. So it can't be related to drop *generation*, because his drops were generated before the save (when he was created). So it is indeed something to do with objects and the floor - maybe there's a bug in floor_carry? Or maybe not ...

d_m: I think this might be my fault. When working on level feelings I refactored object generation so that the feeling is adjusted in the calling function, not in the obj-make code itself. One of the changes was to change make_object from a bool (yes an object was created) to an s32b (the value of the created object). So wherever there used to be if (!make_object...) .... I replaced that with make_object(...); if (!o_ptr->kind) ... do you think you could give those changes a second opinion? They're in 40d63b8.

That description is accurate; I reloaded, killed Smeagol again, and didn't crash.

[Max Stats]

I can confirm that this has happened to me before (game crash as you land the killing blow on a monster). After the panic save, you can reload and kill the monster without incident. The problem is, the only way you could get a savefile for testing is if you knew it was going to happen and got one right before you landed the blow, so reporting the bug is very difficult.

[d_m]

That description is accurate; I reloaded, killed Smeagol again, and didn't crash.

Interesting.

I was musing aloud on IRC last night while looking at this.

It seemed like there *might* be a way for special artifact generation to fail in a way that would create an object with a NULL kind. I will plug that hole and then keep looking.

I played for a few hours last night hoping to trigger the bug (I had GDB ready) but things decided to behave.

[Jazerus]

Tonight's RC build has some improvements to the keymap UI from d_m. In particular, the ctrl-enter bug should be gone: use $ to exit the keymap editor. (If anybody needs to use $ in a keymap, let us know why.)

$ to exit keymap editor does not work for me - it treats $ as a keypress within the keymap rather than an exit command.