03:34:00 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 11:57:09 New branch created: pull/5209 (1 commit) 13https://github.com/crawl/crawl/pull/5209 11:57:10 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5209 * 0.35-a0-291-g12b2ff33e5: Don't spawn batty summons wandering 10(19 minutes ago, 1 file, 4+ 1-) 13https://github.com/crawl/crawl/commit/12b2ff33e515 12:00:26 <09hellmonk> @wizardike do you know if the bug from this run got fixed? https://crawl.project357.org/morgue/Wizard1ke/morgue-Wizard1ke-20260413-010047.txt 13:04:26 New branch created: pull/5210 (1 commit) 13https://github.com/crawl/crawl/pull/5210 13:04:27 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5210 * 0.35-a0-291-gb2b9cc9b68: Fix a bug with exclusions for new enemies on descent 10(17 minutes ago, 1 file, 22+ 9-) 13https://github.com/crawl/crawl/commit/b2b9cc9b683d 13:12:38 <08nicolae> is there a list of which branches naturally have pool_fixup set on their vaults and which don't 13:12:46 <08nicolae> D does but Shoals doesn't, etc. 13:14:42 <09hellmonk> my guess is that it's a branch flag but I don't know for sure 13:27:39 <08nicolae> i figured as much "nicolae can't this wait until you get home" no 13:48:15 <11Odds> Huh, I was totally expecting https://github.com/crawl/crawl/issues/5202 to be wrong, but it looks like we do in fact apply infusion slaying to throwing 13:50:42 <11Odds> Because even throwing attacks have a weapon, so the check https://github.com/crawl/crawl/blob/6208fce01160350e7621a00483614e87f0bd7d33/crawl-ref/source/attack.cc#L865 doesn't work 14:06:15 New branch created: pull/5211 (1 commit) 13https://github.com/crawl/crawl/pull/5211 14:06:16 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5211 * 0.35-a0-291-g1d4bb28395: Fix hurling and don't apply infusion to throwing 10(7 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/1d4bb2839509 14:43:08 <04dracoomega> Oh, oops. I meant to get to that the other day, but got sidetracked. (Throwing attacks didn't used to have a weapon until I refactored ranged attacks to remove the ammo member near the end of 0.34) 15:46:25 <04dracoomega> @Odds So, I have spent a pile more hours running numbers on piety decay. A lot of your math reasonably checks out, though in a handful of cases, the new formulas are giving me consistently inflated piety values at 6* for reasons I've yet to figure out. It could just be random noise or something I'm still overlooking. However, there's a couple definite problems in the math in a few places. There are several examples (eg: for TSO) 15:46:25 of "X accounted for Y% of piety, so we increase piety by Y% to compensate for its removal" but that isn't actually reciprocal. If you remove something that accounded for 33% of piety gained, increasing the remaining piety gain by 33% is still a net nerf of ~12%. You undo a 33% decrease with a 50% increase. There are a few more complications, though. Zin's piety from tithes doesn't show up in the conduct table (since it uses a unique codepath). 15:46:26 However, it also cannot be deduced directly from the piety gain table at 6* since - unlike basically other method of piety gain that I can see - if you're already at 200, you don't tithe at all. So the piety isn't 'eaten' by the cap; it simply never happens. Which actually makes it impossible to verify that the new numbers are correct (for a while, it looked like the new math would result in a passive piety increase at 6* compared to now, but probably 15:46:26 it is being masked by this) 15:46:27 <04dracoomega> Also: I'm not sure how to math out things regarding piety being eaten by gift timers. Like, gift timers eat a fixed percentage of piety gain, so if you are earning less piety gain, they are eating proportionally less. However, their total cost remains the same, so if you are earning less piety, they remain active for longer. And their chance of starting again is based on how much piety you have at the time. Which, all together, 15:46:28 makes me completely unsure what is the correct way to evaluate what a global decrease in piety gain will do to the net piety gained by gifting gods over time. Of course, I kind of wanted to remove the piety cost of gifts altogether, so maybe it would make sense to work on that at the same time? It's sort of funny just how many Sif logs I've looked at now that never hit 6* for the entire game. Did you know that Trog's weapon gifts eat something like 15:46:28 50% of all piety you gain at 4-5*? (Trog earns more piety than most kill gods, due to the anti-wizard conduct, but only by ~12% or so). Removing this would simplify some of this math, but it is a pretty big change to math out on its own. (Trog could probably stand to get a piety income buff, but I don't know that, for instance, Oka needs any such thing) 15:47:40 <04dracoomega> Having looked at just how astronomically high TSO piety gain really is in extended, I don't think I mind the effective reduction in that this does. 15:48:55 <04dracoomega> Really, I have just been consistently surprised by how impactful decay is. It's given me some pause about this whole process, since it seems it really was a tangible reward for playing efficiently and just... nobody really realized it. This isn't necessarily a reason to keep it, but I did go into this originally expecting that it played a much smaller role than it actually did. 15:55:20 I mean I didn't know it existed 16:15:32 <09hellmonk> How were these characters tested btw. I imagine that, due to piety hysteresis, characters that use more god abilities (and therefore spend more time at lower raw piety values) lose less piety to decay. 16:17:51 <09hellmonk> or maybe that is the wrong way to look at it, but I think that parameter is also important for judging total piety available 16:18:40 <08nicolae> sadly i did have to wait till i got home: https://github.com/crawl/crawl/blob/6208fce01160350e7621a00483614e87f0bd7d33/crawl-ref/source/dungeon.cc#L2793 looks like cocytus, swamp, and shoals don't fix up their pools, and also lair looks like it might have a 3/4 chance to call the water fix-up function twice 16:24:11 <04dracoomega> I can't say exactly how Odds arrived at all of their numbers (which do largely seem to be correct) but I have been looking through piles of winning 3 rune characters to double-check various things 16:24:50 <04dracoomega> (And doing math on 4-5* and 6* seperately, due to different stepdowns) 16:24:59 <09hellmonk> I have some more detailed thoughts about this (and clocks in general) but I won't be able to post until tomorrow 16:25:26 <09hellmonk> but I think we have roughly the same perspective and I trust your judgment 16:27:01 <04dracoomega> Well, I'm certainly not pushing anything on this subject tonight 16:28:40 <04dracoomega> Though I would like to settle on what we're doing here in the near future, at least. (And while I'd like to be very clear on the effect of the changes I'm making, there's probably something to be said for how these numbers sometimes varied more widely between gods than I or possibly anyone else fully realized and things were still 'fine' >.>) 16:29:07 <04dracoomega> Certainly, some of the simplifications done in the PR are for complexity that isn't actually buying anything important 16:29:28 <04dracoomega> (I thought the monster holiness differences on piety rate mattered a lot more than they actually did, for instance) 16:29:28 <09hellmonk> Not surprising, I think piety tends to hover at close to max for much of the game unless a character is very weak 16:30:16 <04dracoomega> That seems true less of the time than you might think (or at least like.... probably roughly 40% of characters I've looked at bouncing below 6* again like a half-dozen times or more) 16:31:00 <04dracoomega> And a decent chunk of the characters where this wasn't the case is because they seemed to not be using their abilities >.> 16:31:10 <04dracoomega> Though it definitely varies by god 17:14:01 <07wizardike> No this isn't fixed and it likely affects every version of webtiles with mouse support. Its way more frustrating to exploit in 0.34 and later as you can't move with your turn over anymore 18:13:44 03WizardIke02 07https://github.com/crawl/crawl/pull/5208 * 0.35-a0-291-ga8425ee275: Fix info leaks from feature tile overrides (CrawlOdds) 10(2 days ago, 27 files, 282+ 79-) 13https://github.com/crawl/crawl/commit/a8425ee27579 20:58:21 <11Odds> 1. Interesting re: inflation at 6* - I was all ready to say this might be due to the decision to keep the stepdowns equal, but actually I think that points (very slightly) in the opposite direction. Dunno what's up here. 2. Quite right on the reciprocals (the other important case I see at a glance is Zin's 20% buff). 3. Zin's tithe piety: I think we can get round this by working out the right piety ratio by looking at decay and gain 20:58:21 below 6*. I agree we won't actually be able to verify at six stars, but if we get the basic no-stepdown piety adjustment right that should follow fine. I can reanalyse my Zin logs and see if it matters. 4. On gift timers: I think increasing these proportionally to piety conduct changes works just fine (really thought I'd done this, but maybe I'm remembering a different version). That means that they should tick down in the same length of time. And if 20:58:22 everything else is working, your piety at any time is about the same as it was before, so we shouldn't need to worry about when they start. But also, I agree this would be a great time to remove the piety cost of gifts, if we want to do that (and I'd be keen, I think they are a weird extra complexity). I think if that's happening I'd advocate for a nerf to Oka piety, which frankly is well deserved - his piety gain is lightning fast until gifty time. As 20:58:22 you say, gift piety costs are really significant. Veh is the other god I wonder about here, they have a complicated gifting system and it's also rather load-bearing in character progression (and I've not thought it through). 5. On overall impact - I guess I'm less surprised by you and maybe that's why I did this change rather than someone else 🙂 . But it being higher impact IMO makes removing it more compelling not less; the efficient play it 20:58:23 incentivises is not fun (mostly, it incentivises not autoexploring levels to completion, IMO). 21:01:31 <11Odds> Ability use actually doesn't really matter at all to the maths - we want net piety gain to be the same (at each piety level), which is invariant to abilities. But - for calculating the numbers I used mostly real characters, except where there were gaps (IIRC no-one had played a Zin 3-runer at all). And for my sanity check characters I tried to be somewhat realistic, using abilities as I normally would and only using wizmode to keep me 21:01:32 alive through unusually reckless play. 21:05:33 <04dracoomega> the efficient play it incentivises is not fun (mostly, it incentivises not autoexploring levels to completion, IMO). That may be true. Like, it's easy to imagine what it means is "Bonus for not playing ultra-conservatively and actually moving forward at a reasonable pace without needing to retreat a dozen times" and it's theoretically nice to reward that, but hard to do so without also penalizing things like "Running back to a 21:05:34 shop" or "Autoexploring inefficiently" (which is something that has always plagued larger Crawl clock attempts, I think) 21:06:18 <04dracoomega> Portal vaults are probably the best we have at doing so, I suppose 21:06:55 <11Odds> Yeah I mostly think that the game has done a great job at making soft clocks unnecessary to the point where their downsides clearly outweight heir upsides 21:07:28 <11Odds> (Hard and somewhat stringent clocks like gems and portal vaults are much better IMO) 21:08:19 <04dracoomega> Yeah, I didn't really mean "Oh, piety decay made for an interesting bonus in practice, so maybe removing it is not as compelling now." but was genuinely surprised at the magnitude, which made me examine it in a different light 21:08:29 <11Odds> (My thoughts haven't changed much since I wrote https://www.reddit.com/r/dcss/comments/ra2bs2/piety_decay_is_bad/, though I do somewhat regret writing "Piety decay is easy to replace") 21:08:40 <04dracoomega> Ahahahaha 21:08:46 <04dracoomega> Er, let me explain 21:10:18 <04dracoomega> A couple times over the past couple days I wanted to do a quick check on the wiki's piety decay article and 'dcss piety decay' was autocompleting a google search with the page in question right next to the top, but you know what was actually on the top? And which I saw flash past me like 3 times yesterday? "Some reddit post by someone complaining about piety decay" 21:10:34 <04dracoomega> Literally this one, it turns out 21:11:06 <09hellmonk> piety decay is cringe and reddit confirmed 21:11:16 <04dracoomega> (I mean, it's easy to imagine anyone having complained about it, so I never gave it much mind) 21:12:26 <04dracoomega> But I guess you've actively wanted this gone for a while, while I was mostly motivated by a much quieter "People periodically get confused about losing god powers and is this actually doing anything anyway?" a lot more recently 21:13:12 <04dracoomega> I like how Sif floats to mention in this post, too 21:13:17 <11Odds> Yeah, in a background way I think it's a generally bad thing, and the existence of Sif makes it a somewhat important bad thing 21:13:43 <04dracoomega> Of course, I think gift piety handling is the worst part of that (and is really impactful on them in particular) 21:13:49 <11Odds> If piety decay was a good thing overall, I'd still definitely push for some big sift change 21:14:21 <11Odds> Yeah piety gift handling did surprise me in its magnitude during all this testing even though I already knew it was quite a big deal 21:14:52 <11Odds> Turns out a surprising amount of actualy piety gain for gifting gods happens while your gift timer has run out but the gift hasn't arrived yet 21:15:41 <11Odds> (And Sif gives gifts faster after timeout) 21:23:34 <04dracoomega> But yes, I've been broadly in favor of removing gift timeouts effect on piety for a while (since I knew that one was big and also rather opaque and hard to control on our end.) Much easier to have a knob to independently adjust overall piety to whatever level we want and gift rate to whatever we want. 21:23:44 <04dracoomega> The current situation is basically historical 21:23:50 <04dracoomega> (Just load-bearing for a couple of gods) 21:25:04 <11Odds> Yeah. For Trog and Sif I think there's a strong argument that this should be completely fine, because they are essentially normal piety gods so their systems are battle-tested with many other gods. 21:26:12 <04dracoomega> (I am still on the side of Sif having explore piety for flavor reasons >.>) Though I know gammafunk had reservations. 21:26:18 <11Odds> The thing about Veh is that their gifting system slows down their piety progression a lot and this is probably critical to making the gifts come at good times. Even max piety is probably too early for capstones otherwise? 21:26:45 <04dracoomega> Well, it actually wouldn't be that weird if Vehumet stayed that way, since there's nothing to spend the piety on anyway 21:27:08 <04dracoomega> You inevitably reach 6* eventually and the gifting system isn't really interfering with much 21:27:24 <11Odds> Ah yeah, leaving gift piety for Veh might be a nice compromise 21:28:16 <11Odds> Slightly hard to think of alternatives really, and Veh's pacing is a) quite tricky and b) quite good at the moment IMO 21:28:38 <04dracoomega> Yeah, it feels like that system at least reasonably works as-is, so I don't think it needs much touching 21:29:27 <04dracoomega> Everyone complains about Sif and Trog piety ends up oddly low (which has a pretty significant effect on BiA, since its quality is piety-dependent) 21:30:13 <11Odds> A mild buff to late-Trog wouldn't go amiss IMO 21:34:02 <04dracoomega> On a related (but also somewhat tangential note), a while ago there was some suggestion (was it hellmonk? I don't recall for sure) about Trog letting you throw while berserk and potentially moving ammo gifts from Oka back to Trog. Which I actually found fairly compelling, to be honest; Oka already has more going on in their kit and curtails fewer other options, and Trog feels like they could benefit from an additional point of 21:34:03 interest (and is generally weaker outside of berserkers). I did want to do some larger changes to throwing that I hadn't quite figured out the shape of yet (and thus might not happen for a significant while), but I was somewhat interested in looking into that idea whenever we did so. 21:36:16 <11Odds> I don't remember a time when ammo gifts were on trog 🙂 21:38:35 <11Odds> Sounds sensible and also a big deal for both gods. I kind of wonder if Oka's conduct is a bad thing in that world - a ranged option deals with a lot of situations where allies are another option. (Though also I just dislike the conduct, I think summoners of okawaru were quite a cool niche back in the day) 21:43:53 <04dracoomega> I remember playing multiple needlestabber kobolds of Trog, back in ancient times. (I mean, Oka still gave ammo back then also. They both did. I wonder if there were any specific differences between them, on reflection...) 21:46:01 <04dracoomega> Huh. They apparently only lost it in 0.25 21:46:07 <04dracoomega> I assumed it was further back 21:46:21 <11Odds> Oh, I just my memory failing me then 21:47:30 <02Darby> throwing while berserk sounds like a fairly meaningful buff against blinking enemies, also, allowing one to maintain berserk when they blink away 21:49:58 <04dracoomega> It's true. That part might be... of some concern. But it depends on the shape of throwing at the time (and I do feel like a slightly more diverse throwing system could be something that made the overall Trog kit a bit more interesting, and fit well) 21:51:51 <11Odds> On a different and even more tangential note - how would you feel about an option that made autoexploration biased to stay near upstairs? (The reason this is even tangential is I expect such an option would cause a mild slowdown in exploration, but if piety decay isn't a thing this matters less) 22:01:29 <04dracoomega> I am fairly neutral on that. It feels like a thing some people would like having, at least. 22:01:51 <04dracoomega> Don't we already have a couple of different autoexplore bias options? I vaguely feel there were a couple. 22:03:02 <11Odds> There's just one, and it's about wall-hugging 22:04:22 <04dracoomega> (I have never myself used, nor investigated, any of this, to be clear) 22:04:37 <11Odds> If it's not an obvious won't-do I'll probably give it a go and see how it turns out in practise 22:05:29 <04dracoomega> I can't see any reason why it would be negative 22:05:47 <04dracoomega> Assuming it's not wildly slow or something (I don't see why it would be) 22:08:10 <11Odds> Re: oka. It looks like he achieves 4* piety at least twice as fast as a regular god. IDK if we still want him to get there faster after gift piety changes or not? As a god with a very solid 1* ability I don't see any compelling reason for this fast start. Though I guess he does technically remove some power at 0* 22:12:40 <04dracoomega> I actually wasn't even aware that this was the case (and it might often escape attention given that he's definitely not especially fast to hit 6*, which is the most talked about threshold in his case) 22:13:00 <04dracoomega> Though I guess it's true that I often end up with Finesse before I have even slightly reasonably trained the invo to use it 22:14:36 <11Odds> This fast start definitely matches my experience 22:14:47 <11Odds> Particularly I always notice heroism feels almost immediate 22:17:50 <04dracoomega> In theory, Oka is supposed to reward you extra for taking extra-hard fights, but I wonder how well that actually works out in practice. Like, it does have a psychologically positive impact when you get the Big Piety, but I wonder if it actually changes what anybody does? (It doesn't have to, mind you) 22:18:08 <11Odds> It changes what I do, occasionally 22:18:09 <04dracoomega> But I am just thinking of the interaction between this piety speed and the fact that earlygame is full of red stuff compared to later 22:18:42 <11Odds> I spend resources to take on a fight I can win (e.g. curare on an ogre I could let sleep) 22:19:38 <11Odds> Ah yeah, this is a good point. So actually oka piety probably doubly slows at the moment - naturally and because of gifts 22:21:23 <11Odds> (If it didn't already exist, I'd be pretty unsure about whether the nice flavour oka having a separate piety system is worth the complexity) 22:23:05 <04dracoomega> Perhaps. (Though it's pretty well-encapsulated and easier to reason about than gift piety is >.>) 22:35:42 <11Odds> BTW I'm assuming you are going to do the changes implicit in the conversation above (deal with gifts somehow, fix the reciprocals thing) as part of merging this, rather than me add to the PR. Happy to do it the other way if it would help. 22:42:23 <04dracoomega> There's some you could do with fixing the logging to differentiate piety lost to stepdowns from piety that just doesn't do anything since you already have 200, and displaying Zin tithes in the piety gain table (along with fixing the berserker issue) for later verification, but it probably is better if just one of us handles changes to the piety PR at a time and I already have my nose in it. 22:43:05 <04dracoomega> (Not sure how long it will take me to get the gifting piety stuff removed. At this rate I'm going to guess "Longer than I'd like" ^^; ) 22:44:27 <11Odds> Cool, I’ll have a look at just that logging stuff 23:47:09 Monster database of master branch on crawl.develz.org updated to: 0.35-a0-290-g6208fce011