00:14:26 -!- amalloy_ is now known as amalloy 01:21:28 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-3222-g5327be9f7f 01:33:04 Unstable branch on crawl.develz.org updated to: 0.26-a0-72-g09a6a6ccbd (34) 03:30:41 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-3222-g5327be9f7f 03:40:03 Fork (bcadrencrawl) on crawl.kelbi.org updated to: 0.22.1-2783-g2ab4c1891b 05:27:24 -!- amalloy is now known as amalloy_ 06:01:18 -!- Amnesiac_ is now known as Amnesiac 07:04:40 -!- amalloy_ is now known as amalloy 11:35:40 hm, I don't understand https://crawl.develz.org/tournament/0.25/clan-exploration.html 11:36:10 how are the scores so large if it is still distinct runes and distinct branch entry/end milestones? 11:36:27 compare with https://crawl.develz.org/tournament/0.25/exploration.html 11:59:52 seems like it must be summing the unique points for each team member 12:00:11 just a guess from the totals 12:00:26 er, just summing, not unique points 12:00:55 it's not actually a sum, but the totals seem proportional to what the sum would be 12:02:45 hm, it is supposed to be the sum 12:05:30 nm I don't understand either 12:08:17 also not sure these views are really working as intended for unclanned players 12:08:29 but hopefully that doesn't impact any actual scoring 12:08:44 the clan scoring views have a WHERE team_captain IS NOT NULL 12:08:54 so they should be excluding unclanned players 12:09:12 some of the intermediate ones don't, but yeah hopefully doesn't matter 12:09:17 e.g. clan_scaled_rune_find_count 12:14:06 clan scaled_rune_Find_count is the culprit 12:14:12 it's using group by player instead of team captain 12:14:30 ugh 12:15:11 the fix is to pause taildb, drop and recreate the view, and then restarit taildb 12:17:49 some day we need to build more help into this system for players to set a team name 12:31:05 are there reasons why we require TEAMCAPTAIN to be on the first line, and in upper case (or at least, it looks like it checks case to me on a quick scan)? 12:31:31 I just did some grepping after helping someone on discord and there are 25 players with it not on the first line, and 8 with it in lowercase 12:36:28 no one chose camel case at least 12:52:39 ebering I can do that at some point 12:53:29 can you also commit a fix for database.sql? 12:54:02 I wonder if I'll have to drop the views that depend on it too 13:06:19 Unstable branch on crawl.akrasiac.org updated to: 0.26-a0-72-g09a6a6c (34) 13:10:52 I think we require first line to make it easier to troubleshoot multiple rcfile stuff 13:11:20 in case someone has something buried somewhere in an old rcfile 13:11:27 don't think that part is unreasonable at all really 13:12:27 uppercase is definitely just historical reasons 13:13:07 (both the first line requirement and the uppercase requirement predate my involvement with the scripts - the only thing I added there was support for multiple servers) 13:14:16 I think first line also makes the TEAMNAME/TEAMMEMBERS setup simpler to parse for captain rcfiles 13:14:26 since that's really two lines etc 13:14:33 yeah, I was considering whether to drop it for players only 13:14:45 I suspect either way we end up troubleshooting 13:16:15 I think no servers copy rc files across versions any more? 13:16:17 some used to 13:17:19 if that's the case then that should cut down on issues with rcfile collision at least 13:17:36 looks like a lot of these sorted it out and just didn't delete the wrong line, so maybe 25 is an overestimate 13:17:57 as p0werm0de pointed out, it does very clearly say first line 13:18:10 someday we can have a textbox like it's 2005 13:18:32 might drop the case requirements though? 13:19:20 dropping case requirements sounds good 13:29:51 Inescapable altar vault 13https://crawl.develz.org/mantis/view.php?id=12278 by Midn8 13:34:19 advil: you won't I don't think? and yes 13:34:38 I'm trying to win this FoFE first to get the nem points 14:13:41 so the good news is that this salt cloud inconsistency bug is pretty trivial to repro 14:13:48 Clouds only affect enemies in desolation of salt 13https://crawl.develz.org/mantis/view.php?id=12279 by FakerFangirl 14:14:17 wander into desolation, find some salt clouds, save and reload a few times 14:14:20 you'll end up with a different set of salt clouds 14:14:25 that is... not the correct description for that bug 14:16:55 hm, mantis seems very slow... anyway, it seems to happen anywhere that cloud generators exist. looks like more clouds get generated on each reload 14:17:04 (with the usual random chance) 14:17:50 PleasingFungus is it salt only? or all clouds? 14:17:58 Volcano too 14:18:03 did you test from scroll of fog? 14:18:10 it might be a load level bug 14:18:12 ooh, good question. i'd bet *against* scroll of fog 14:18:16 let's see 14:18:16 there's some dodgy stuff with clouds in there 14:18:31 but yeah it could be that something gets reset with the generators 14:18:42 scroll of fog seems unaffected 14:19:47 there's also dodgy stuff in there with markers 14:20:14 it's not entirely impossible that this is the sort of thing that would have gotten confused when I refactored level loading with pregen 14:26:59 ok, this seems fairly straightforward to explain 14:27:12 whenever we enter a level, we fire DET_ENTERED_LEVEL, which triggers fog machines 14:27:20 *whenever we load a level 14:28:05 advil: I've pushed the corrected schema 14:28:20 but don't reload the whole db, you should be able to just migrate the view 14:28:26 since views don't have constraint checking 14:29:28 Hello dear devs, I have a 0.25 tournament game on CXC that crashes everytime I enter Snake 4. This seems pretty exploitable: when I load the game back, I spawn on the bottom of the stairs and the floor is reinitialized, so I can reliably refarm it to get uber loot. 14:30:01 ventricule: can you post a save backup, please? 14:30:09 how do I do that? 14:30:12 it's online 14:30:13 https://crawl.develz.org/learndb/#save_backup 14:30:46 hm, i wonder if I have admin on cxo 14:31:21 you can go to http://crawl.develz.org/wordpress/howto and look at the section for CXC to see how to log in via SSH 14:31:26 (if you're a tiles player) 14:32:09 sorry, I know it's some extra steps, but it shouldn't be too much 14:32:19 I'm on it 14:32:23 thanks! 14:32:46 re the cloud stuff earlier: looks like fog machines are the only lua affected by this. timed messages (i.e. portal vaults) and other timers are also triggered by the entered_level event, but they do handle it correctly 14:33:55 here's a better link re save backups: https://crawl.develz.org/wiki/doku.php?id=mantis:savebackup 14:34:12 thanks! sorry, i'm out of date... 14:34:38 uh oh 14:34:45 uh oh? 14:34:47 fixing this chei banner will require a db reload 14:34:57 or changing the description to the old one 14:35:02 since that's whats programmed in 14:35:09 wait maybe not 14:35:12 :) 14:35:16 no, it will 14:35:23 :( 14:35:37 well 14:35:40 the old chei banner is nice 14:36:16 I guess I'll change the description and then for 0.26 we can get it right 14:36:16 https://crawl.xtahua.com/saves/ventricule-crawl-0.25-200614-1835.tar.bz2 14:36:33 thanks! 14:38:48 hm, i can't unzip the save. advil, what am i doing wrong? 14:39:03 $ bzip2 -d ventricule-crawl-0.25-200614-1835.tar.bz2 bzip2: ventricule-crawl-0.25-200614-1835.tar.bz2 is not a bzip2 file. 14:40:16 it worked for me 14:40:35 well, with whatever is doing it automatically on download on a mac 14:41:46 not sure why but there is also a file named `ventricule-crawl-0.25-200614-1835.tar.bz2.cpgz` that showed up 14:42:44 oh hm 14:42:55 that save I just gave you is not the right one I think 14:43:00 it seems to be from 2019 14:44:32 the tar.bz2 file is only 2692 bytes so there's no way it's a dcss save 14:45:29 PleasingFungus haha indeed, if you look at the contents ot that file it is the save downloader cgi-bin 14:45:42 so I think save downloading must be broken on cxc 14:47:16 crashlog is here: https://crawl.xtahua.com/crawl/morgue/ventricule/crash-ventricule-20200614-182734.txt 14:52:45 I can replicate it without the save, so no need to get it 14:54:22 Let me know if you want me to experiment with other staircases or anything 14:55:59 thanks, it looks like this is just crashing on any levelgen for snake 4 (annoyingly on a pass that is vetoed, so it's not even relevant to what ends up generating) 14:59:45 ventricule2 this is going to take a couple minutes, sorry, unfortunately I think your save will not be playable without intervention 15:08:42 No problem, can I still finish this game (and skip s4) or should I just start another one? 15:12:25 because of pregen, I think it will crash any time you descend below a certain depth 15:13:00 I think I have figured out the problem, so I can fix it soonish (and it won't impact seeds thankfully) 15:31:10 Ok, thanks for the hard work! 15:31:15 hm, is there any way to print from lua? 15:31:39 nvm, got it 15:37:58 New branch created: cloud_gen_reload (1 commit) 13https://github.com/crawl/crawl/tree/cloud_gen_reload 15:37:58 03PleasingFungus02 07[cloud_gen_reload] * 0.26-a0-73-gabc7bfb: Fix cloud generators triggering on load 10(56 seconds ago, 3 files, 11+ 4-) 13https://github.com/crawl/crawl/commit/abc7bfb7dccb 15:38:58 advil: ^ want to eyeball this before i merge? 15:39:48 i very strongly suspect this bug is over a decade old 15:40:54 %git 45b0ff6c34bfa8f1653413349a31a65070e662c6 15:40:54 07Matthew_Cline02 * 0.4-a0-252-g45b0ff6: Added fog machine Lua map markers. Needs a C interface to make it usable in level generation. 10(13 years ago, 15 files, 471+ 64-) 13https://github.com/crawl/crawl/commit/45b0ff6c34bf 15:41:05 PleasingFungus stupid question maybe, why didn't you just add LOAD_RESTART_GAME to the conditions that prevent that event from firing? 15:41:28 not a stupid question 15:41:57 I think it's a good behavior that the event triggers timed portal messages 15:41:59 well, hm 15:42:01 maybe that's a bug too 15:42:02 let me think 15:42:19 if it gives distance messages on reload, that's a bug too, and we should just shut it off entirely 15:42:34 heh I hadn't actually realized that was for all events 15:42:41 so my *premise* was stupid 15:42:59 no, it just gives "you hear the sound of rushing watter" 15:43:01 *water 15:43:08 i mean, this is used in very few places 15:43:54 it's cloud generators, timed triggers, and timed messages 15:43:54 one of those is only used for portal vaults (I think), and I'm not sure what the other one was used for 15:44:13 oh, sorry, missed your PM earlier - sounds like you have it under control, though? 15:44:27 yeah 15:46:01 given this your fix seems good for now but I kind of want to say that the API would be better if DET_ENTERED_LEVEL were not called on reload, and DET_LOADED_LEVEL (or something) were 15:46:29 at least your change makes it much more obvious that that event would fire on a level load 15:46:39 hopefully more obvious 15:46:49 i'll merge it for now. do i backport it into 0.25? i forget the policy 15:46:58 yeah, I think it's important enough to 15:47:26 Branch master updated to be equal with cloud_gen_reload: 13https://github.com/crawl/crawl/tree/master 15:47:48 we have done major bugfixes at least for the tourneys I've been involved in 15:47:48 so i checkout stone_soup-0.25 and push the fix to that, right? 15:47:52 maybe also credit the reporter? 15:48:23 oops... 15:48:23 agree 15:48:36 how can this be a major bugfix if it is over a decade old? 15:48:36 do we allow force pushes to fix my bad history? or is that too messy now that it's on master 15:48:37 guess you can edit the commit msg for 0.25 at least...yes, cherry-pick -x to stone_soup-0.25 15:48:42 heh 15:48:46 no force pushes to master 15:48:49 i mean, it's a pretty bad bug even if no one noticed 15:49:11 if any servers happen to pull before you do it, it requires manual intervention on their end 15:49:16 right, ok 15:49:21 also force pushes on master are fundamentally scary 15:49:47 elliptic: would you prefer that i not cherry-pick it? 15:49:56 no, I'm fine with it being cherry-picked 15:50:59 just a little confused about what the definition is of major :P 15:51:38 03PleasingFungus02 07[stone_soup-0.25] * 0.25.0-2-g7881a98: Fix cloud gen triggers on load (WindshieldViper) 10(15 minutes ago, 3 files, 11+ 4-) 13https://github.com/crawl/crawl/commit/7881a984e165 15:51:57 "bad enough that we would be willing to push during tourney" :-P 15:52:08 I guess it's exploitable now that people know about it if they get into trouble near a cloud generator 15:52:30 yeah, that was the specific case that led to it being reported 15:52:37 quite strong in desolation 15:53:03 the player was running away and repeatedly reloading until LOS was blocked 15:53:53 The build has errored. (cloud_gen_reload - abc7bfb #13614 : Nicholas Feinberg): https://travis-ci.org/crawl/crawl/builds/698289566 16:01:37 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-73-gabc7bfb7dc (34) 16:02:57 oops, sorry, deleted the branch midway through CI 16:11:39 lol my debugging of this turns out to have been greatly compounded by the fact that there were like 8 VETOs and it tried to place minmay_shoptagon on at least two of these passes 16:19:26 what's the hypothesis? 16:19:34 the hypothesis is that advil two weeks ago left a comment being "what's up with this assumption I am making" and that was a good question 16:19:40 haha 16:21:29 there's a bunch of code that, when an unrand is called for by but has already generated, produces a somewhat similar randart instead. For staves, it wasn't handling the case where the artifact data didn't set a fallback subtype, and it tried to generate a random one, instead leaving OBJ_RANDOM as the subtype 16:30:48 ohh and this is a staff generated in a backhand way via weapons 16:30:57 no wonder I'm confused 16:31:05 (and no wonder the item itself is confused) 16:39:50 -!- jfcaron_ is now known as jfcaron 16:45:56 wow 16:46:56 -!- jfcaron_ is now known as jfcaron 16:49:07 03advil02 07* 0.26-a0-74-g64415b1: Fix some crashes with fallback randarts for staff of battle 10(6 minutes ago, 1 file, 32+ 16-) 13https://github.com/crawl/crawl/commit/64415b183b01 16:49:07 03advil02 07* 0.26-a0-75-ga645b5f: Minor tweaks to levelgen item debug msgs 10(2 hours ago, 2 files, 4+ 4-) 13https://github.com/crawl/crawl/commit/a645b5fb1833 16:49:45 03advil02 07[stone_soup-0.25] * 0.25.0-3-gd7cc568: Fix some crashes with fallback randarts for staff of battle 10(7 minutes ago, 1 file, 32+ 16-) 13https://github.com/crawl/crawl/commit/d7cc5688da20 16:50:27 scary 16:52:51 it's kind of messy 16:56:46 I wonder what the best way would be to request a stable rebuild on cxc these days 17:01:43 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-75-ga645b5fb18 (34) 17:17:45 hm did someone force push new-scoring? 17:17:58 something weird happened to a cosmetic commit I made 17:18:27 oh maybe I just never pushed 17:26:44 !version 17:26:49 trunk: 0.26-a0-72-g09a6a6c; 0.25: 0.25.0-3-gd7cc568; 0.24: 0.24.1-14-gf8d2b50; 0.23: 0.23.3-3-gaaff5ab; 0.22: 0.22.3-3-g34f245f; 0.21: 0.21.3-3-g218456f; 0.20: 0.20.2-3-g29329e0; 0.19: 0.19.6-3-g7d5932f; 0.18: 0.18.2-3-g5ac21b8; 0.17: 0.17.3-4-g93a25c6; 0.16: 0.16.3-4-gc506b73; 0.15: 0.15.3-3-g8e88d0d; 0.14: 0.14.3-3-g02348ce; 0.13: 0.13.3-3-g4e9f7b7; 0.12: 0.12.4-3-g8567966; 0.11: 0.11.4-3-g4668e67; .. 17:31:52 ebering ok the clan exploration table should be fixed 17:32:01 new values look more reasonable 17:32:08 and elliptic 17:34:59 -!- Tux[Qyou] is now known as goat 17:35:06 -!- goat is now known as Tux[Qyou] 17:40:41 haha someone just asked if it was possible to revisit the acquirement option list after reading the scroll and dropping it somewhere 17:41:01 I guess the next logical step would be to goldify revisiting it into a one-use ability 17:41:09 (which might not actually be horrible ui-wise!) 18:13:48 lol well this might provide clues to the excursion bugs: https://github.com/crawl/crawl/issues/1454 18:14:11 "wanted to check downstairs [with X], and got cards [from nemelex] when i did so." 18:22:10 advil: great 18:26:52 Unstable branch on underhound.eu updated to: 0.26-a0-75-ga645b5fb18 (34) 19:14:48 -!- amalloy is now known as amalloy_ 19:29:26 -!- Tux[Qyou] is now known as Thugs[Qyou] 19:44:25 interesting, the overview page lists the top 5 clans as all rank 1, then 6,7,8,9,10 for the next 5 as one would expect 19:44:36 (the full clan ranking is correct) 19:44:47 they're all winners! 20:22:20 Stable (0.25) branch on underhound.eu updated to: 0.25.0-3-gd7cc5688da 20:24:32 https://docs.google.com/document/d/19iVmxa3m5qwuqqI2vt8sdk4oMslu719XBfZ3jcGhR6M/edit?usp=sharing i'm not sure if what i'm describing in this doc will ever land in mainline non-forked crawl, but i wanted to write it up anyway 20:24:43 some 'food for thought, as they say 20:29:53 I don't think I agree with the assumption here that things changing over the course of the game is bad? 20:31:10 like you are treating this as a given but that seems wrong to me 20:31:41 among other things, it's worth noting that player accuracy can go down as well as up at various points in the game 20:31:47 it's not some monotonic thing 20:32:30 if you put on heavy armor, switch to a bigger weapon, or switch weapon classes to something with less skill, your accuracy will go down 20:33:06 while raising your fighting or weapon skill or dexterity will make your accuracy go up 20:33:56 I would interpret the current accuracy/monster EV system as designed to make these changes noticeable without making them too restrictive 20:36:50 i'm not sure i took it as a given... i thought I tried to explain why this particular change seemed questionable 20:37:25 so your hitrate against a given "high-EV" monster might start at 50% or something (making up numbers) and end at 80% or something, but it will go up or down at various times bouncing back and forth 20:37:36 sure 20:37:50 i guess the flipside question is, why we want this change? why do we want it to go from 50% to 80%, or whatever? 20:37:52 so there's a sense of change without having a risk of people's build landing them in a spot where their hitrate is painfully low 20:38:33 it feels strange to me that we have this enormously complex system that mostly doesn't matter 20:38:41 it's also worth noting that there's still a ton of different between different monsters 20:38:54 yes, i did note and approve of that! 20:39:24 you note that a bat and a caustic shrike have similar EV, but throughout the game there are monsters with 0 EV and 7 EV and 15 EV and they feel substantially different 20:39:50 that's why i gave examples of both 'high-ev' and 'typical' monsters - I also mentioned 'low-ev monsters' like hydras in an earlier section 20:40:19 basically I think you need a much clearer reason why the current system is flawed 20:40:51 before proposing something like your 2 at the end 20:40:51 players have all these strategic decisions (like weapon type, skilling, armour) that impact accuracy, but it's almost never correct to make those choices based on their impact on accuracy 20:40:51 ? 20:40:55 that's just not true 20:41:05 do you ever level fighting to increase your accuracy? 20:41:37 that's definitely part of it early on, and I'd say the main reason I raise armor skill early on is to help accuracy in plate 20:41:56 maybe i'm way off base. the only time i've ever heard players mention those options is to ridicule them 20:42:07 and to specifically warn newer players away from making decisions based on accuracy 20:42:43 like lots of things affect melee damage 20:42:50 things that affect accuracy are one of them 20:43:40 I'm not saying "armor skill is better than weapon skill because of accuracy" 20:44:44 idk, probably i'm not communicating well... 20:44:47 but I'm saying "armor skill is worth raising, if to a lower level than weapon skill, and accuracy is the main reason for that early on" 20:45:22 interesting. i haven't done the math on that, and i'm not sure how i would, but i believe you that could be correct 20:45:24 fwiw, I agree that putting the % chance of hitting a monster in the monster description is a good idea 20:46:20 i wonder if shields and rmsl (on deep elf archers?) cause issues for that idea 20:46:20 to-hit is *relatively* straightforward otherwise... 20:46:32 I just don't think that there's a dichotomy between "early game high-miss-rate" and "late game high-hit-rate" 20:47:42 as in, the great majority of characters will have higher accuracy at the end than the start, but there's a lot that I'm glossing over in the middle? 20:47:47 I think it's a continuum that you move between and that things go up and down, and that having some mixture depending on details of the character is probably good 20:49:17 i also like that things go up and down, i'm just not sure about the destination 20:49:42 the destination is that you get to the enchantress and miss a lot of times :P 20:49:47 @??the enchantress 20:49:47 the Enchantress (13i) | Spd: 10 (move: 60%) | HD: 15 | HP: 84-117 | AC/EV: 10/30 | Dam: 26 | 10weapons, 10items, 10doors, spellcaster, see invisible | Res: 06magic(160) | XP: 3895 | Sp: slow, dimension anchor, mass confusion, strip resistance, haste, repel missiles, invisibility [04emergency] | Sz: little | Int: human. 20:49:52 a fine destination 20:50:02 also worth noting that oofs have 20 EV 20:50:04 (i also didn't mention OOFs) 20:50:05 yes 20:50:26 didn't think it was very different from 18 EV, but maybe i should've swapped them in for caustic shrikes 20:50:31 despite the iconic nature of the acid bee 20:51:03 @??executioner 20:51:03 Executioner (151) | Spd: 20 | HD: 12 | HP: 57-77 | AC/EV: 10/15 | Dam: 30, 10, 10 | 05demonic, 10doors, fighter, evil, see invisible | Res: 06magic(140), 05fire, 02cold, 10elec, 03poison, 04rot, 13neg+++, 13torm | Vul: 08holy | XP: 2325 | Sp: pain (d14) [06!sil], haste [06!sil] | Sz: Large | Int: human. 20:51:11 monster AC also doesn't scale that much fwiw outside of extended I think? if you recognize that relatively early monsters like orc warriors can be wearing plate 20:51:51 @??orc warrior ; plate armour 20:51:51 orc warrior (08o) | Spd: 10 | HD: 4 | HP: 22-32 | AC/EV: 9/10 | Dam: 20 | 10weapons, 10items, 10doors, fighter | Res: 06magic(20) | XP: 132 | Sz: Medium | Int: human. 20:51:51 @??orc warrior 20:51:51 orc warrior (08o) | Spd: 10 | HD: 4 | HP: 23-32 | AC/EV: 8/10 | Dam: 20 | 10weapons, 10items, 10doors, fighter | Res: 06magic(20) | XP: 132 | Sz: Medium | Int: human. 20:51:51 hm, that doesn't seem right 20:51:51 @??orc warrior ; robe 20:51:51 orc warrior (08o) | Spd: 10 | HD: 4 | HP: 22-32 | AC/EV: 1/13 | Dam: 20 | 10weapons, 10items, 10doors, fighter | Res: 06magic(20) | XP: 132 | Sz: Medium | Int: human. 20:51:51 wow, weird 20:51:55 i guess monster defaults to thinking they have chain? 20:52:11 I forget exactly how monster AC works 20:52:18 note that 9 AC in plate is unexpected too 20:52:28 what would you expect? 20:52:29 since plate is 10 AC base 20:52:34 heh, true 20:52:35 I would expect at least 10 :P 20:52:56 @??skeletal warrior 20:52:56 skeletal warrior (10z) | Spd: 10 | HD: 10 | HP: 52-76 | AC/EV: 15/10 | Dam: 25 | 07undead, 10weapons, 10items, 10doors, fighter, evil, unbreathing | Res: 06magic(100), 02cold, 09poison+++, 12drown, 04rot++, 13neg+++, 13torm | Vul: 08holy | XP: 771 | Sz: Medium | Int: human. 20:53:24 it might be a fun project to take objstat results and graph, eh, maybe some histograms? % of level that has < 5 EV, 10 EV, 15 EV etc for each generated level 20:53:38 should be doable if you ignore monster equipment 20:53:53 and then likewise for AC 20:55:27 anyway, thanks for talking it through with me! it was really helpful 20:55:44 monster shields are also maybe something that haven't entered into this discussion enough 20:55:58 how much should they enter this discussion? 20:56:09 since they effectively act as giving some monsters super-EV 20:56:10 @??mennas 20:56:10 Mennas (11A) | Spd: 15 | HD: 19 | HP: 133-175 | AC/EV: 15/26 | Dam: 30, 20 | 08holy, 10weapons, 10items, 10doors, fighter, spellcaster, see invisible, fly | Res: 06magic(160), 10elec, 03poison, 04rot, 13neg+++, 08holy+++ | XP: 7878 | Sp: mass confusion, silence, minor healing (2d9+3) | Sz: Medium | Int: human. 20:56:13 they do 20:56:40 oh, bad idea for a weapon class gimmick: what if maces ignore sh 20:57:20 (or halve it or something, you know) 20:57:34 don't think enough monsters have shields for that to be a good idea, yes 20:57:46 clearly a better gimmick for monsters attacking players?! 20:58:14 I look forward to seeing "orc warrior (mace/flail)" on my monster lists 20:58:20 :) 20:59:13 anyway I do think your first recommendation of showing hit chance in descriptions of monsters makes sense, maybe in the same line as the current EV bar 20:59:44 i've had some energy for crawl projects later, so maybe i'll put out a patch... 21:00:10 what did you think about the idea of showing it in the targeter for spells/ranged attacks? too noisy? 21:00:31 I guess maybe one last note is that functionally "You miss the monster" and "You hit the monster but do 0 damage" are identical 21:00:55 it's true! 21:03:07 I guess in summary I feel like the model of "early game melee and late game melee feel different to the player because of accuracy" is too simplified 21:05:30 showing accuracy in the targeter for spells/ranged attacks seems probably fine for most things 21:06:08 if there are spells with weird targeters where you really want to target a square other than the one the monster is on, that could be a bit weird 21:06:46 there are also some weird questions with bolt spells that lose 1 range for each monster they hit 21:07:11 if that is still a thing? (I'm not sure I believe it should be) 21:07:17 i have no idea 21:07:26 wouldn't it be like... only bolt of magma? 21:08:25 so if you are firing a bolt at a yak at the end of your range and there's another yak between you, and your spell's accuracy would give 80% chance to hit a yak 21:08:46 then you'd only have a 16% chance of hitting the yak at end of range :P 21:09:11 !! 21:09:31 hm, let me just do a quick check of beam.cc... 21:09:42 (probably you don't actually want to try to do this calculation in the display, but it's amusing) 21:10:51 i think it's only bolt of magma 21:11:40 important drawback! 21:46:49 -!- amalloy_ is now known as amalloy 21:58:33 It's nice that you can use morgues to see how frequent misses are today 21:58:44 and per-XL too 21:59:33 maybe the first step in any accuracy reform would be to build a system that approximates the current one but using a more transparent set of calculations. Then it would be easy to make experimental adjustments and see how gamefeel changes 22:07:16 hm, what would make accuracy more transparent? 22:07:57 right now, we have a to-hit, which is *mostly* constant (with some big exceptions), and a target EV, which we already display (as pips) 22:08:12 how would we simplify that further? 22:18:36 I mean, the discussion above about how you sometimes want to level armour to increase accuracy, sometimes fighting, sometimes weapon skill... it's easy to show the player "you will hit bat 60% of the time.", but how is that info actionable currently? 22:19:46 there are so many factors that go into accuracy calculation. And ultimately it's just one component of the whole melee attack system. The complexity feels out of keeping with its relative importance. 22:19:54 it seems a lot less complex than, say, damage calculation 22:21:15 i think the idea was that you could use the accuracy percentage numbers to (a) compare weapons ("i wield the exec axe and have a 30% to-hit, vs 70% with the hand axe...") and (b) evaluate tactics ("the komodo dragon is 'almost dead', and i have a 60% to-hit...") 22:22:31 ok, those are two great ways to action this info. I was thinking about the case (c) increase accuracy ("my current acuracy is too low, which of n ways to increase it should I pick?") 22:23:47 b in particular is very great IMO. It would lead to much better transparency in those "do I attack this ogre or use a consumable" situations that new players (& me) pick wrongly in 22:23:56 this approach certainly has limitations - it can't really show the impact of skill 22:24:03 (well, not in advance) 22:24:18 and showing it so prominently might lead to people overweighting it 22:25:16 it seems worth trying, though... and might expose some weird behaviors we currently aren't aware of 22:26:54 accuracy is by far the simplest factor in melee damage so worrying about simplifying it seems weird to me 22:27:50 "by far the simplest" != "simple"!! 22:28:01 it can actually fit in a learndb entry 22:28:27 ( https://crawl.develz.org/learndb/#accuracy , rip sequell :( ) 22:30:05 one way in which the info is useful is for stuff like "how much did my accuracy just go down when I switched from ring mail to plate mail with 17 strength and 2 armor skill" 22:31:21 it also would give a better idea how how EV differences between monsters actually affect hitrate 22:31:27 PleasingFungus: technically that's spread over several entries of accuracy and also test_hit 22:31:41 technically! 22:31:57 it also gives you an idea of how much extra fun "inacc provides :) 22:32:01 I appreciate it's not that useful to be shown numbers with no context but we certainly thought about examining a monster just showing you the chance to hit with your current weapon. 22:32:06 in general I'm much happier about displaying numbers like hitrate or failure rate that are meaningful without additional info 22:33:06 i think if you are talking about accuracy reform, it's a good chance to remove a bunch of factors from that calculation. For example weapon enchant & slay could affect only damage, hunger penalties don't need to exist any more, the invis/halo/umbra could be standardised 22:34:00 what does "standardised" mean? 22:34:16 "the same (absolute) number" 22:34:24 but why is that better? 22:34:33 that number isn't intrinsically meaningful 22:35:04 it makes the formula shorter and easier to understand 22:35:13 is that important for players? 22:35:16 but why do players need to understand the formula, yes 22:36:15 I think it's important because players already reimplement this calculation externally (eg the anydice link). So there is a desire to mentally understand the formulas 22:36:15 I would MUCH rather have a somewhat more complicated formula that plays better than one that is easier to compute by hand 22:36:34 I expect 99% of players don't "reimplement this calculation externally" 22:36:36 well, I don't think making all those numbers the same would really play worse 22:38:26 did someone really come along in the past and go "hmm, umbra should affect accuracy less than halo because "? 22:38:32 probably? 22:38:46 like these numbers matter 22:39:16 it's pretty rude IMO to just assume that whoever implemented umbra didn't worry at all about balance and just picked random numbers that have no significance 22:39:33 you don't design around making formulas more elegant 22:42:32 I'm sorry, you're right. It's not clear to me where the accuracy numbers for all these small constants came from, or why they were picked. I tried searching umbra's numbers specifically but they seem to be there from pre-git days. 22:43:04 wait, I thought umbra was newer than that? 22:43:13 I don't think elegance of formula should be given greater weight than gamefeel 22:43:33 gamefeel! 22:43:44 my assumption about random numbers is that someone picked them at some point and then people playtested the effect for years 22:43:48 now you're speaking my language 22:43:52 and if people decided to tweak them, they did 22:44:20 I don't think you can balance anything by pure formulawork regardless of how simple the formulas are 22:45:09 What I am trying to say, but I think I was talking around this, was that I agree that most of the time, increasing formula clarity would reduce gamefeel 22:45:34 however, it could be possible to substantially simplify a formula for a minimal (perhaps undetectable) reduce in gamefeel 22:45:35 596895e2a0 added umbra btw 22:47:15 (0.10) 22:47:45 huh weird, the git log I was grepping was incomplete 22:58:14 (doesn't look like it was tweaked any time in the last decade, fwiw) 22:58:47 (but probably it's better to be careful about reducing to-hit, since that's a much more potentially powerful/unfun effect than increasing to-hit) 22:59:11 oh, elliptic, did you see me talking about the korean fork that adds an "omniscience" power to Ash? 22:59:24 did not see 22:59:56 for a short duration, all of your attacks autohit and all enemy attacks automiss (if applicable) 23:00:11 doesn't feel very Ashenzari to me, but it's a cool mechanical gimmick 23:00:36 that's the fork that has a beogh family tree 23:01:49 https://namu.wiki/w/%EB%8D%98%EC%A0%84%20%ED%81%AC%EB%A1%A4/%EC%B6%94%EA%B0%80%EC%A3%BD#s-3.7.4 tell me this isn't fantastic... 23:04:11 I think calling that a family tree might be a little inaccurate, but giving the unique orcs new titles when they level up is cute 23:06:51 autohit/automiss invocation is sort of neat, agree that it's not very Ash but could be cool somewhere at some point 23:08:29 In _acquirement_weapon_subtype() there's a line specifically to consider formacid's shield skill as 0 in order to make sure they always weight towards 2Hers...this line doesn't actually do anything, because it checks handedness based off of the player's handedness, thus everything is 1Her for Formacid anyways. 23:11:28 It's minor. I guess I'll put it in as a bug report with the source cleanup tag with recommendations. 23:12:09 PleasingFungus: maybe wu jian heavenly storm 23:16:04 03advil02 07* 0.26-a0-76-gbe553b7: Fix timing for map view cursor clamping 10(4 hours ago, 1 file, 6+ 3-) 13https://github.com/crawl/crawl/commit/be553b73f188 23:16:04 03advil02 07* 0.26-a0-77-g726680d: Don't call redraw_screen while off-level 10(10 minutes ago, 1 file, 151+ 144-) 13https://github.com/crawl/crawl/commit/726680df4ae0 23:16:04 03advil02 07* 0.26-a0-78-gd6c187f: Crash if show_init is called off-level 10(2 minutes ago, 1 file, 2+ 0-) 13https://github.com/crawl/crawl/commit/d6c187ff355b 23:17:35 so uh it's possible that the redraw fix there should be merged to 0.25 too 23:17:38 it's a bit crazy that map knowledge gets updated by viewwindow 23:18:33 alexjurkiewicz: yeah maybe! i still haven't played a wu or usk game (I think)... 23:26:57 There's a line in _acquirement_weapon_subtype() which has no effect. 13https://crawl.develz.org/mantis/view.php?id=12280 by bcadren 23:28:14 advil: that is an intense bug 23:29:57 it's very crawlcode 23:30:45 still a bit aghast that that the ui redraw call is what causes map knowledge to be updated when you move 23:31:09